chore: release v0.10.0

[avrabe]

Canonical-ABI correctness sweep — 15 fixes since v0.9.0, all driven by the mythos-auto delta-pass on #179. Bumps meld-core and meld-cli from 0.9.0 → 0.10.0.

Headline (memory-safety / priority high)

• LS-P-6 params/return-area accumulator wrap → OOB write in cabi_realloc (now saturating).
• LS-P-7 conditional-pointer CopyLayout per-leaf instead of composite-payload (7/8 under-copy for list<u64> inside option/result).
• LS-P-8 Record/Tuple field-walk used unpadded child size at 25 sites; spec-correct now uses canonical_abi_element_size.
• LS-P-9 total_flat_params saturating fold (was sum::<u32>); fixes calling-convention swap.
• LS-P-10 nested-conditional outer-guard chain on ConditionalPointerPair — closes an arbitrary-cross-component-memory-read with attacker-controlled (ptr, len) for result<option<string>, u32> etc.
• LS-P-14 nested-list inner-copy buf_len overflow guard.
• LS-P-16 UTF-16→UTF-8 lone-high-surrogate OOB read; conservative trap (U+FFFD upgrade is follow-up).
• LS-P-19 UTF-8→UTF-16 mirror: each multi-byte branch guards continuation-byte reads.

Defensive mitigations / hardening

• LS-P-12 + LS-P-18 refuse list<option/result/variant-with-pointer> (silent stale-pointer corruption); full per-element conditional fixup is follow-up.
• LS-P-11 flat-name resolver rejects duplicate module exports.
• LS-P-13 async adapter param-copy uses resolver positions, not the (i32,i32) heuristic.
• LS-P-15 OOB resource_type_id dropped + warned, not silently misclassified.
• LS-P-17 warn on mixed-encoding caller before the heuristic fallback miscompiles string_transcoding.

Hygiene

• flat_byte_size element-wise variant JOIN (was max; no in-tree consumers, no LS-N).

Tests / verification

• 266 lib tests (+34 from v0.9.0's 232)
• LS-N gate: 33/33 approved loss scenarios with regression tests
• mythos-auto delta-pass: clean per-file on parser.rs / fact.rs / resolver.rs
• Each finding clean-room-verified by an independent agent following validate.md before fix-or-dispute. One false positive (auto-runner disputed LS-P-8's spec-correct fix, directly contradicting canonical-abi.py::record_size) was disputed rather than fixed.

🤖 Generated with Claude Code

[github-actions]

LS-N verification gate

✅ 33/33 approved LS entries verified

	count
Passed (≥1 test, all green)	33
Failed (≥1 test failure)	0
Missing (no ls_*_NN_* test found)	0

_{Approved loss-scenarios.yaml entries are expected to have a
regression test named ls_<letter>_<num>_* (e.g. LS-A-11 →
ls_a_11_*). The gate runs each prefix via cargo test --lib --no-fail-fast and aggregates pass/fail/missing.}

Failed LS entries

(none)

Missing regression tests

(none)

_{Updated automatically by tools/post_verification_comment.py.
Source of truth: safety/stpa/loss-scenarios.yaml.}