[tlwillke]

Fixes #779.

NVQParameters.idStringFor passed two arguments to a three-placeholder format string ("NVQ_%s_%d_%s"), so it threw MissingFormatArgumentException. The trailing _%s is removed, so the cache key is now NVQ_<dataset>_<nSubVectors>, matching the PQ key's style.

Nothing constructs NVQParameters today, so behavior doesn't change; this removes a latent failure for when NVQ is wired up as a compressor option.

Testing: adds CompressorParametersTest, which checks the NVQ cache key.

[github-actions]

Before you submit for review:

• Does your PR follow guidelines from CONTRIBUTIONS.md?
• Did you summarize what this PR does clearly and concisely?
• Did you include performance data for changes which may be performance impacting?
• Did you include useful docs for any user-facing changes or features?
• Did you include useful javadocs for developer oriented changes, explaining new concepts or key changes?
• Did you rebase your branch onto the latest main for regression testing and PR submission?
• Did you trigger regression testing via Run Bench Main and review results?
• Did you adhere to the code formatting guidelines (TBD)
• Did you group your changes for easy review, providing meaningful descriptions for each commit?
• Did you ensure that all files contain the correct copyright header?
• Did you add documentation for this feature to the release notes directory?

If you did not complete any of these, then please explain below.

[MarkWolters]

LGTM

[MarkWolters]

Cached NVQ compressors would be reloaded as PQ (fixed in #776)

The key fix here is right, but it makes NVQ caching reachable, and the load path can't read what it saves. With idStringFor working, supportsCaching() == true takes effect: Grid.getCompressor saves the NVQuantization to pq_cache/NVQ_<ds>_<n>. On the next run, Grid.loadFromCache reads every cache file with ProductQuantization.load. NVQ's layout starts with a version int, which PQ reads as a v0 centroid length, so you'd get an exception or a corrupt ProductQuantization where an NVQuantization was expected. The test here only checks the key string, never a save and reload, so it doesn't catch this.

loadFromCache is code that #776 (mine) moves and rewrites, so I've fixed it there rather than ask for it here:

• Adds CompressorParameters.loadCompressor(RandomAccessReader). The default throws UnsupportedOperationException, like idStringFor. PQParameters uses ProductQuantization.load and NVQParameters uses NVQuantization.load.
• Grid.loadFromCache now calls cp.loadCompressor(...) instead of always ProductQuantization.load.
• Includes the same idStringFor change as this PR ("NVQ_%s_%d"), because NVQ caching can't be exercised without it.
• Adds TestCompressorCaching, which round-trips PQ and NVQ through loadCompressor and checks the NVQ key.

How the two interact:

• Either order merges cleanly. The idStringFor line is identical in both, and your test is a new file under jvector-examples/src/test, separate from the new one in Refactor grid #776.
• No broken window in between. Nothing constructs NVQParameters outside tests yet, and the YAML compression types don't include NVQ. So if this merges first, the PQ-load path can't be reached before Refactor grid #776 lands.
• Your test's supportsCaching() assertion is accurate once Refactor grid #776 is in.

So this can merge as is, or you can close it in favour of #776. Either works for me.

[r-devulap]

LGTM.