fix(rust): a module path names the module's free function, not a method

[danusha2345]

Problem

A Rust call written through a module path — crate::util::take(3), super::util::take(3), self::util::take(3), or util::take(3) after use crate::util; — maps the module prefix to its file correctly, then takes the first node of that name in the file. method was one of the accepted kinds, and a function local to another function's body was not ruled out. So a module that declares impl Buf { pub fn take(&self, …) } (or trait Take { fn take(&self); }) above its free pub fn take(…) sent every path call to the method; a module with only the method bound the call to it. use crate::util::take; landed on the method too and was then dropped, leaving no import edge.

Fix

In resolveRustPathReference, a Rust path never names a method (that is Type::method) nor a function-local item: drop method from the accepted kinds and keep only lexically reachable items (isLexicallyReachable, the same check #2224 applies to the framework heuristics). A module whose file declares no such item still falls through to name matching, which links nothing.

Tests

__tests__/rust-module-path-call.test.ts (real temp Cargo project + SQLite): crate/super/self/module-after-use paths, a use of the function, a nested a::b path, a module with only a same-named method, and a module not in the index. All 4 cases fail on main and pass with the fix (kernel and wasm extraction). Full suite on Linux passes.

A/B (calls edges, main → fix)

• A GNSS simulator crate: 13 calls retargeted from GnssTimeConverter::utc_to_gps_time (and 6 siblings) to the module's free wrapper functions of the same name, plus 7 use crate::gnsstime::{…} import edges; every call site is a crate::gnsstime::<fn>(…) path, so 13/13 and 7/7 are correct.
• Five other Rust crates: no change. Node counts unchanged everywhere.

Not in this PR

use crate::util::take; take(3) with another free take elsewhere (e.g. other.rs) still goes to the other one: the uses.bound set from #2214 keeps leaf names only, not the module the use names. Happy to follow up separately.

🤖 Generated with Claude Code

[danusha2345]

Folded into #2308, which carries this PR's source change and its whole test file (plus an associated-type decoy case) and adds exact resolution of bare calls through their use binding on top. Closing this one so the two don't overlap.