[Burapa-J]

Fixes #3762

Cause

1. own_calls (compiler/crates/hy-check/src/walk.rs:545-549) only matched top-level PlaceKind::Index, skipping calls inside PlaceKind::Field (arr[mutate(m)].val = 99), PlaceKind::AddrOf, and nested PlaceKind::Index (mat[mutate(m)][0] = 99) targets.
2. own_fnlits (compiler/crates/hy-check/src/walk.rs:618-622) only traversed values on Stmt::Assign, skipping Expr::FnLit closures inside assignment target index expressions.

Fix

• Add recursive place_calls and place_fnlits helpers in compiler/crates/hy-check/src/walk.rs mirroring place_uses.
• Call place_calls in own_calls and place_fnlits in own_fnlits for all Stmt::Assign targets.
• Add 4 integration tests in compiler/crates/hy-check/tests/walk_assign_target_place_calls.rs.

Verification

• cargo test -p hy-check --test walk_assign_target_place_calls (4 passed)
• cargo test -p hy-check --lib (183 passed)

[limitlessprojects]

Landed — thank you @Burapa-J. main now includes this PR as merge commit 44af2b93 (trial-tree-asserted d473df3a — the tree differs from your head 956fe0a9 only because the merge carries main's newer state including #3761, and the full battery ran on exactly that merged tree — remote-asserted).

What landed (fixes #3762 — Stmt::Assign target places only collected calls from bare Index targets, so calls/closures nested inside field and addr-of target projections were invisible to the statement's own-calls/own-fnlits attribution):

• New place_calls / place_fnlits walkers descend target places recursively (Index base+index, Field base, AddrOf base), wired into own_calls and own_fnlits.
• #123 lineage preserved: closure bodies are still handled as units by the caller — the A5 perimeter keeps flatten/uses/calls outside closure bodies.

Verification evidence (full battery on the merged trial tree):

• merge gate: clean vs main 4ccbd0b7
• cargo test --workspace: 452 suites ok, 2924 tests, 0 failed (your walk_assign_target_place_calls.rs: +117 lines, all green)
• corpus harness: 135/135 verdict agreement, tripwire not fired
• differential FULL: accept 45/45, reject 90/90 — FULL AGREEMENT

Bounty rail: per BOUNTY.md / discussion #295.