CNDB-15554: Bump jvector version

[michaeljmarshall]

What is the issue

Fixes: https://github.com/riptano/cndb/issues/15554

What does this PR fix and why was it fixed

Bring in the latest jvector changes. Including a bug fix for a regression introduced in rc.3

Commits: datastax/jvector@4.0.0-rc.3...4.0.0-rc.5

[michaeljmarshall]

As it relates to test coverage, there are several code paths that are only reachable if you already have a fused adc graph already or already have bq vectors. At the moment, we don't have a way to write either of these objects/formats, so there isn't a way to add test coverage for them. #2042 will add fused adc and therefore add coverage. BQ is on the road map for later this year. As such, I am going to merge even though we have less than the required test coverage.

As it relates to the CNDB PR, we already did the release validation testing with this version on the october release, so I am going to merge and then open the CNDB PR to bring in the latest CC.

[michaeljmarshall]

Just discovered I can't merge with the current code coverage, so I guess I'll have to look to see if there is a way to fix this.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[cassci-bot]

❌ Build ds-cassandra-pr-gate/PR-2033 rejected by Butler

---

2 regressions found
See build details here

---

Found 2 new test failures

Test	Explanation	Runs	Upstream
o.a.c.index.sai.cql.VectorBruteforceLocalTest.multiSSTablesTest[ed] (compression)	REGRESSION	🔴🔵	0 / 11
o.a.c.net.ProxyHandlerConnectionsTest.suddenDisconnect (compression)	REGRESSION	🔴🔵	0 / 11

---

Found 1 known test failures

[michaeljmarshall]

the failure seems flaky:

Expecting actual:
  0.75
to be greater than or equal to:
  0.8

It passed on my machine.