refactor(server): respect cgroup heap limits

[imbajin]

Purpose of the PR

Server launchers size the default heap from host memory, so a memory-limited container can receive a heap larger than its limit. The default calculation now uses the smaller of host availability and remaining memory under visible cgroup hard limits.

Main Changes

Resolve covering cgroup v1/v2 mounts and their visible ancestor limits. Account for reclaimable inactive file cache, using conservative values when statistics are missing or inconsistent. Preserve existing heap bounds and explicit JVM option precedence.

Run License Checker before the Server product plans and required Memory tests in one workflow. Keep both required check names and isolate the Memory plan artifact. Preserve limited automatic retries for required jobs on failed push runs, using the native job rerun API; successful unrelated product jobs are retained.

Verifying these changes

• Need tests and can be verified as follows:
  • Cgroup fixtures, ShellCheck, existing launcher GC checks, CI policy tests and actionlint passed.
  • Three independent final reviews passed; live Linux warm-cache pressure remains to be validated.
  • Formatting and clean compilation passed with Java 17.
  • A standard Server container limited to 4 GiB started and restarted with the patched sizing helper.

Does this PR potentially affect the following parts?

• Dependencies
• Modify configurations
• The public API
• Other affects: default Server heap sizing in memory-limited Linux deployments.
• Nope

Documentation Status

• Doc - TODO
• Doc - Done: Docker Server heap sizing.
• Doc - No Need

Heap sizing at a glance

The diagram illustrates the default calculation in this change. Explicit heap options retain precedence.
The heap is only part of process memory; native allocations, stacks, and caches still need headroom.
This illustration describes the design, not runtime acceptance or merge readiness.

[imbajin]

Independent review of this head: 7.0/10; hold for the two P2 fixes below, then re-review. The supplied shell fixture suite and bash syntax check passed independently. Two additional bounded cloud fixtures reproduced the decision paths described inline. No real container restart, memory-pressure test, full Java build, or RC acceptance was run in this review. The author's container validation remains separate evidence. CI was still running when reviewed.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 52.51%. Comparing base (6aa362a) to head (919aa62).
⚠️ Report is 2 commits behind head on master.

Additional details and impacted files

@@             Coverage Diff              @@
##             master    #3320      +/-   ##
============================================
+ Coverage     52.42%   52.51%   +0.09%     
- Complexity     2273     2374     +101     
============================================
  Files           762      762              
  Lines         67400    67431      +31     
  Branches       9138     9151      +13     
============================================
+ Hits          35333    35410      +77     
+ Misses        28525    28488      -37     
+ Partials       3542     3533       -9     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.