Cache resolveAsset calls to improve web performance

[ezraripps]

Caches graph asset path / metadata construction internally with resolveAssets to improve performance on web.

Web calling resolveAssets 190 times goes from 6.5s to ~50ms now with the cache addition.

---

Note

Medium Risk
Touches core asset-resolution and reachability logic; fast vs FX-like result paths could diverge if railInfo aggregation doesn't match the old scan behavior.

Overview
Caches graph-invariant lookup structures for AnchorGraph.resolveAssets (and listAssets, which delegates to it) so repeated filtered calls—e.g. one per source asset on web—no longer rebuild keys, adjacency buckets, and rail aggregation on every invocation.

Adds a memoized #getResolveIndex() that builds a ResolveIndex once (node keys, rail-inclusive join maps, asset-location buckets, and pre-aggregated railInfo). BFS now seeds from computeStartIndices, using bucket lookup when both asset and location are set instead of scanning the full node list. Result assembly uses buildResultMapFast from railInfo when onlyAllowFXLike is false; the FX-like path still scans nodes to exclude non–FX-like edges.

^{Reviewed by Cursor Bugbot for commit 6bce448. Bugbot is set up for automated code reviews on this repo. Configure here.}

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
88.1% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud