Fix flaky Scala CI Bloop timeouts

[schani]

Summary

• give Scala CLI's Bloop health checks an additional two minutes under fixture load
• limit Scala fixture jobs to two concurrent workers to reduce compilation-server contention
• retain the Bloop warmup because direct compilation is substantially slower and overflows Scala 3.2.2's compiler stack on large generated sources
• pin the Scala CLI setup action to v1.15.0

Why

Scala fixture jobs intermittently fail while querying the shared Bloop compilation server:

java.util.concurrent.TimeoutException: Future timed out after [30 seconds]
    at bloop.rifle.BloopRifle$.getCurrentBloopVersion(...)

This is an infrastructure failure rather than a generated-code failure. Scala CLI's SCALA_CLI_EXTRA_TIMEOUT extends the hard-coded 30-second Bloop health-check timeout, while reducing fixture concurrency makes it less likely that Bloop is too busy compiling to respond.

Example failure: https://github.com/glideapps/quicktype/actions/runs/30034893445/job/89327258609?pr=3046

Verification

• npm run lint
• PATH="/tmp:$PATH" CPUs=1 QUICKTEST=true FIXTURE=scala3 npm run test:fixtures -- test/inputs/json/samples/reddit.json
• PATH="/tmp:$PATH" CPUs=2 QUICKTEST=true FIXTURE=scala3,scala3-upickle npm run test:fixtures -- test/inputs/json/samples/reddit.json test/inputs/json/samples/pokedex.json

[github-actions]

No generated-output differences

✅ This PR does not change generated outputs.

[github-actions]

No generated-output differences

✅ This PR does not change generated outputs.