fix: restore support for declare option exist:timeout "-1" to disable timeout

[joewiz]

Summary

• Restores the ability to disable query timeouts via declare option exist:timeout "-1" (or "0")
• setTimeoutFromOption() was missing the <= 0 guard that configureDefaults() already had, causing negative values to be stored directly as the timeout — since elapsed > -1 is always true, queries were immediately terminated
• Adds XQueryWatchDogTest with 7 unit tests covering positive, zero, negative, invalid, and multi-value timeout options

Fixes #2529

Test plan

• New XQueryWatchDogTest passes (7/7 tests)
• Verify declare option exist:timeout "-1"; no longer causes immediate query termination in a running eXist-db instance

🤖 Generated with Claude Code

[joewiz]

Ubuntu CI timeout analysis

The ubuntu-latest unit (Java 21) job has been timing out after 45 minutes on repeated runs of this PR. After investigation, we suspect this is a pre-existing infrastructure issue unrelated to the PR's changes.

Why we suspect this PR isn't the cause:

1. The PR changes are trivially small: a 3-line if guard in XQueryWatchDog.java and a new unit test file with 7 simple EasyMock-based tests. Nothing that could cause a hang.
2. The previous run of this exact PR (run 22698232712) had ubuntu unit tests pass in ~16 minutes.
3. The unit test step runs mvnd test across all modules, so any pre-existing flaky test anywhere in the project can cause the hang.

Why only ubuntu is affected:

The CI matrix excludes Windows and macOS from the unit test type — they only run integration tests. So only ubuntu would exhibit a unit test timeout.

Likely culprit:

Known flaky infrastructure tests like MoveResourceTest and RenameCollectionTest can hang indefinitely on CI runners. Re-running the failed job should resolve the issue.

We welcome other ideas about what might be causing this timeout.

🤖 Co-authored by Claude Code

[duncdrum]

We should put a hold on this one, until the open xsuite PRs are in. The hang is a concern, imv.

[joewiz]

@duncdrum Sure, makes sense! I've changed this PR to draft to make it clear that we shouldn't merge it in this state.

The other PR of mine showing ubuntu failures despite repeated runs is #6083.