Fix VectorSessionFunctions element grow/shrink corruption (VectorSet Attributes) + tests

[tiagonapoli]

What

Fixes a VectorSet element corruption bug in VectorSessionFunctions.InPlaceWriter, and adds a unit test over the element sub-store write path that pins it.

The bug

InPlaceWriter did newValue.CopyTo(value) straight into the existing record's value span with no capacity or content-length handling. For a variable-length per-element value (the SETATTR Attributes blob), an in-place overwrite in the mutable region would:

• Grow (e.g. 9 -> 11 bytes) -> ArgumentException: Destination is too short.
• Shrink (e.g. 11 -> 9 bytes) -> stale trailing bytes remain and the content length is never shortened, so the read returns the old length.

The base contract computes a RecordSizeInfo and returns false when the new value won't fit, letting Tsavorite allocate a fresh record — the Vector override skipped all of it.

The fix

Mirror the main-store pattern: build the RecordSizeInfo for the new value, PopulateRecordSizeInfo, then TrySetContentLengths, which resizes/shortens the value in place and returns false when it cannot grow within the allocated record. On false, return false so Tsavorite falls back to a fresh, correctly-sized record via InitialWriter. Only copy once the length is set.

var sizeInfo = new RecordSizeInfo() { FieldInfo = GetUpsertFieldInfo(logRecord, newValue, ref input) };
functionsState.storeWrapper.store.Log.PopulateRecordSizeInfo(ref sizeInfo);
if (!logRecord.TrySetContentLengths(sizeInfo.FieldInfo.ValueSize, in sizeInfo))
    return false;   // -> Tsavorite RCU -> InitialWriter (fresh record)

Test

VectorElementUpsertResizePreservesValue([Values(false, true)] inReadOnlyRegion) drives a dedicated Vector session directly (mirroring VectorManager's DiskANN write callback) and walks a single element through initial write -> equal-size overwrite -> grow -> shrink, verifying the exact round-trip after each stage. It covers both the mutable region (InPlaceWriter) and the read-only region (fresh InitialWriter record). Plus VectorElementUpsertRoundTrips as a baseline.

Before the fix, the mutable-region case failed at the grow stage; with the fix both regions pass.

Validation

• New VectorElement* tests: 6/6 pass (both TestFixtures).
• Full Garnet.test.vectorset suite: 143 passed, 22 skipped, 0 failed — no regressions.

Scope note

Only the Upsert/InPlaceWriter path is fixed here — that is the path attributes actually use (WriteCallbackUnmanaged -> Upsert) and the one the repro hit. The RMW/Updater callback path (InitialUpdater/InPlaceUpdater/CopyUpdater) has the same latent shape but is only invoked by DiskANN with fixed-size data and isn't reachable variable-size through any command, so it is intentionally left unchanged.