migrate to bigint backed (u)int64

[hackwaly]

Note: This PR needs upcoming compiler release to merge.

Summary

This PR migrates the JS-side Int64/UInt64 implementation to a BigInt-backed representation and removes the old manual { hi, lo } emulation path.

Concretely, this change:

• switches JS 64-bit integer handling to use native BigInt semantics
• collapses the split int64_js.mbt / int64_nonjs.mbt implementation into a single intrinsic-based builtin/int64.mbt
• updates JS interop helpers such as BigInt::from_int64, BigInt::from_uint64, Int64::to_string, UInt64::to_string, and
  env.now to match the new representation
• replaces the custom JS Double -> (U)Int64 conversion logic with saturating runtime intrinsics
• reuses the same double_to_int64 implementation for JS, Wasm, and Wasm GC
• removes obsolete target-specific files and simplifies builtin/moon.pkg target selection

Motivation

The main goal is to stop maintaining a separate hi/lo-based 64-bit emulation path on JS and align the implementation more closely
with the intrinsic model used on other backends.

This reduces target-specific duplication, simplifies the runtime boundary, and makes the behavior of 64-bit integer operations
easier to maintain going forward.

Compatibility

No intended public API changes.

There are also no .mbti changes in this branch, so this should be a refactoring from the package interface perspective.

Testing

• Not run locally yet.

---

[coveralls]

Pull Request Test Coverage Report for Build 3720

Details

• 1 of 1 (100.0%) changed or added relevant line in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 95.006%

---

Totals
Change from base Build 3718:	0.0%
Covered Lines:	14610
Relevant Lines:	15378

---

💛 - Coveralls

[devin-ai-integration]

✅ Devin Review: No Issues Found

Devin Review analyzed this PR and found no potential bugs to report.

View in Devin Review to see 5 additional findings.