Refactor build parallelism computation into a dedicated script

[coderabbitai]

Summary

Track a follow-up refactor to move build parallelism computation into a dedicated script instead of routing Taskfile evaluation through components/core/tools/scripts/utils/build-and-run-unit-tests.py.

Rationale

The current implementation works, but the parallelism calculation logic is being reused outside the unit-test utility. Keeping that computation in a dedicated script should make the ownership clearer, reduce coupling between unrelated build paths, and make future updates easier.

Affected areas

• taskfile.yaml
• components/core/tools/scripts/utils/build-and-run-unit-tests.py
• Any shared build helper introduced for computing max parallel jobs

Proposed changes

• Extract the max-parallelism calculation into a dedicated script with a narrow, reusable interface.
• Update Taskfile and other build entry points to call the dedicated script instead of the unit-test utility.
• Keep the precedence rules and behaviour unchanged unless a deliberate follow-up change is proposed.

Acceptance criteria

• A dedicated script is the single source of truth for computing C++ build parallelism.
• taskfile.yaml no longer depends on build-and-run-unit-tests.py for G_CPP_MAX_PARALLELISM_PER_BUILD_TASK.
• Existing override behaviour for CLP_CPP_MAX_PARALLELISM_PER_BUILD_TASK is preserved.
• Existing memory-aware parallelism behaviour remains covered by validation or tests.

Backlinks

• Follow-up from PR fix(build): Cap C++ compilation parallelism by memory limits. #2348: fix(build): Cap C++ compilation parallelism by memory limits. #2348
• Review discussion: fix(build): Cap C++ compilation parallelism by memory limits. #2348 (comment)

Requested by @jackluo923.