perf(explore): searchSymbols binary-insert + complete the total order

[justrach]

Second perf round, item 1 (smallest, highest-certainty from the 4-agent speedup survey).

What

searchSymbols' appendOne kept its K-bounded candidate buffer sorted by re-sorting the whole buffer on every accepted insert — O(K·log K) compares per insert with the default K=50/cap 200 — plus a redundant final sort. Since the buffer is always sorted, placement is a binary search + single insert (O(log K) compares, O(K) moves).

The comparator also gains line_start as the final tiebreak. Without it, same-name symbols in the same file (multiple nested inits, overload-style patterns) compared equal, so their relative order still fell out of hash-map iteration order — the exact residual nondeterminism the deterministic-ordering rewrite (#641) meant to eliminate. With the key ending in (path, line) and same-(path, line) deduped, the order is total and the sorted-insert invariant is airtight.

Evidence

• A/B on this repo (fresh servers, exact/prefix/pattern/fuzzy/kind filters, max_results up to 200): payload bytes identical on every query; embedded dispatch timings: warm prefix:"handle" 3.1ms → 896µs, name:"main" 1.2ms → 931µs.
• Gated bench: codedb_symbol +0.23% (the 21-file corpus exact-name case barely touches the insert path — the win scales with match count, i.e. prefix/fuzzy/glob on real indexes).
• zig build test: 23/23 steps, 869/873 (4 platform skips) including a new same-name-same-file ordering test. e2e: 20/20.

🤖 Generated with Claude Code

[github-actions]

Benchmark Regression Report

Thresholds: 10.00% and 50,000 ns absolute delta

NOISE means the percentage threshold was exceeded, but the absolute delta was too small to fail CI.

Tool	Base (ns)	Head (ns)	Delta	Abs Delta (ns)	Status
codedb_bundle	70918	73864	+4.15%	+2946	OK
codedb_changes	6695	7008	+4.68%	+313	OK
codedb_context	281075	260538	-7.31%	-20537	OK
codedb_deps	297	298	+0.34%	+1	OK
codedb_edit	44599	45435	+1.87%	+836	OK
codedb_find	1937	1747	-9.81%	-190	OK
codedb_hot	14333	14873	+3.77%	+540	OK
codedb_outline	10524	11583	+10.06%	+1059	NOISE
codedb_read	9377	10928	+16.54%	+1551	NOISE
codedb_search	64049	70909	+10.71%	+6860	NOISE
codedb_snapshot	73646	73735	+0.12%	+89	OK
codedb_status	4940	5488	+11.09%	+548	NOISE
codedb_symbol	47623	45316	-4.84%	-2307	OK
codedb_tree	26297	26219	-0.30%	-78	OK
codedb_word	6933	8889	+28.21%	+1956	NOISE