[ci] Bound the runtime of the integration-test jobs

[weiqingy]

Linked issue: #889

Purpose of change

None of the jobs in ci.yml set timeout-minutes, so a hung test runs until GitHub's six-hour per-job ceiling before it is killed. A recent run consumed 6h0m16s on a stalled it-python job — the local llama-server segfaulted mid-run, pytest hung, and the job burned shared Apache runner capacity for a result that was knowable in minutes.

This bounds the three jobs that install and exercise a local Ollama server (it-python, it-java, cross-language), which are the ones exposed to that hang. Each limit is roughly two to three times the job's slowest observed run over recent green builds on main:

job	slowest observed (p100, 6 green main runs)	timeout-minutes
it-python	26.2m	60
cross-language	24.8m	60
it-java	14.0m	40

The segfault itself is a separate question tracked in #889; this change only ensures such a hang fails fast instead of running to the six-hour ceiling.

Tests

No test changes. Verified that the workflow YAML still parses and that the three timeout-minutes keys attach at job level to python_it_tests, java_it_tests, and cross_language_tests. The behavior — a job aborting once it exceeds its limit — is a GitHub Actions primitive.

API

No public API changes.

Documentation

• doc-not-needed

[weiqingy]

The 2 failing checks are unrelated to this change — this PR only adds timeout-minutes to three jobs, so it changes how fast a hang surfaces, not whether any test passes. The Ollama install succeeds in both jobs (>>> Install complete).

• it-java [java-21] [flink-2.0]: the local llama-server crashes (io.github.ollama4j.exceptions.OllamaException: llama-server process has terminated in YamlLoaderIntegrationTest / ChatModelIntegrationTest). This is the model-server crash — intermittent on main too (this job passes 6/6 recent runs). Log.
• it-python [python-3.11] [flink-1.20]: a py4j/Flink-1.20 classpath failure (org does not exist in the JVM, KeyError: 'p'), not Ollama-related. The failure count is identical to the same job on [ci] Retry Ollama install and propagate download failures #891 (which doesn't touch ci.yml), and the same job fails identically on current main (main run), so it's pre-existing and flaky. Log.

The llama-server segfault is exactly the intermittent hang this PR's timeouts are meant to bound (rather than fix) — the root-cause discussion is the open item on #889.

[weiqingy]

Rebased onto current main and squashed out the empty re-trigger commits, so this is now a single commit against a base that includes #896.

One correction to the CI triage above: the it-python py4j failure was described there as an environmental / base failure. It was not — it was an in-repo test bug (the MCP e2e test forking the PyFlink process and corrupting the Py4J gateway), now fixed by #896. It remains true that it was never caused by this PR. Details on #893.

That leg should therefore be clean on this run. The llama-server segfault is unaffected by #896 and can still redden a leg at random; that one is tracked in #893.

[wenjin272]

LGTM