…able-js into alexvictoor-no-constructor-returns

…2072)

* Call super()" before this in constructors

* Fix issue with MapIterator re-setting the next method

…wsers might re-compile immutable

Replace buble by babel

Define "exports" in package.json

The base docstrings were carried over with the mixin migration, but the
kind-specific overrides have their own wording in immutable.d.ts
(Keyed: Object form + key stringification; Indexed/Set: Array form).

- concatFactory: the base collection type is statically iterable since
  Symbol.iterator moved out of the mixin onto CollectionImpl, so the
  stale comment is rewritten and the `as unknown as` bridge becomes a
  plain downcast (the remaining lie is only the `unknown` element type).
- entrySeq: the patched `fromEntrySeq` round-trip is a plain downcast
  too (KeyedSeqImpl subtypes the SeqImpl returned by toSeq); the TODO
  documenting the indexed-source lie stays.
- ToIndexedSequence: replace the `?? 0` fallbacks with non-null
  assertions plus a comment, stating the ensureSize invariant instead of
  masking it.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WvEjriPc3Rbx3sqApa2coD

Only `this.includes` is off-limits (it would recurse through Set's own
`includes` → `has` delegation); `super.includes` is exactly the wanted
semantics and avoids duplicating the base method body.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WvEjriPc3Rbx3sqApa2coD

Unconditional assignment creates own properties whose value is
`undefined` when the first iterable is unbranded. Today the predicates
test truthiness so this is invisible, but the `in`-based checks planned
for 6.0 (see the isOrdered @ts-expect-error note) would read such a
property as branded, making every ConcatSeq look keyed, indexed and
ordered at once.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WvEjriPc3Rbx3sqApa2coD

The `& { [IS_ORDERED_SYMBOL]: true }` intersection was repeated across
the three guard signatures; a named OrderedBrand alias carries the
rationale once and keeps the signatures readable.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01WvEjriPc3Rbx3sqApa2coD

…es-to-ts

Migrate operations/sequences.js to TypeScript and move the order-safe mixin methods onto the classes

The mixin entries in CollectionImpl.js only existed because Collection.ts
cannot import operations/sequences.ts without a cycle. Installing the
methods from sequences.ts itself keeps them next to the wrapping sequence
classes they build, fully typed, and shrinks the mixin further.

Collection.ts now declares the public types: fromEntrySeq joins the
throwing placeholders and concat gets its base d.ts signature, which also
makes the existing *SeqImpl.concat declare-narrowings valid against a real
(bivariant) base method.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EJWDpKjoLcWomxKD3WEmvd

Snapshot of migrated vs remaining files, the mixin dismantling state, the
TODO [TS-MIGRATION] inventory, the source-pass type-test skip list, and
the findings of a behavioural JS->TS consistency review (recorded here,
to be addressed in follow-up PRs).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EJWDpKjoLcWomxKD3WEmvd

Per review on the follow-up PR: the kind narrowings from the d.ts land
with the migration itself instead of a deferred TODO. The three
*CollectionImpl classes get declare properties (type-only, valid against
the base bivariant method), and SeqImpl redeclares concat as a real
method for the public Seq contract — a declare property there would
break the structural *SeqImpl -> SeqImpl assignability — overwritten at
runtime by operations/sequences.ts like the base one, via a new
Seq-typed concatFactory overload.

Un-skips the 'Set concat' source-pass type test.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EJWDpKjoLcWomxKD3WEmvd

The end goal is to emit the public .d.ts from the sources, so migrated
types must stay as close as possible to immutable.d.ts and may only
deviate in the stricter direction (tighter types, or new type
parameters replacing unknown/any). Refresh the status report
accordingly.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EJWDpKjoLcWomxKD3WEmvd

…igration-bctlmg

Move sequence methods from mixin and add TS-migration status report

countByFactory/groupByFactory keep their runtime bodies; the Map they
build is typed through the d.ts import until Map.js is migrated. The
countBy/groupBy mixin entries move onto the prototype from
aggregations.ts itself — same pattern as operations/sequences.ts — with
throwing placeholders carrying the public d.ts types on CollectionImpl.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EJWDpKjoLcWomxKD3WEmvd

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EJWDpKjoLcWomxKD3WEmvd

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01EJWDpKjoLcWomxKD3WEmvd

…gations-to-ts

Migrate operations/aggregations.js to TypeScript

Fix some migration issue reported by AI

Each finding from .agents/migration-status.md was traced back to its
originating PR, then fixed or confirmed as an intentional change:

- R1 (isSubset string argument, from #2204): keep the new
  character-iteration semantics, consistent with isSuperset and with
  Collection('abc'); documented as a v6 breaking change in the CHANGELOG
  and pinned by multi-char unit tests.
- R2 ('@@iterator' fallback removal, from #2127): confirmed intended
  modernisation; explicit CHANGELOG entry, and the two immutable.d.ts
  docstrings no longer mention @@iterator. immutable.js.flow is left
  untouched since Flow's @@iterator syntax denotes Symbol.iterator.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0122shfP2ngkkjuXZBWoeFGW

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0122shfP2ngkkjuXZBWoeFGW

…-yzlr2p

Resolve the three remaining likely-bugs findings (R1-R4)