fix(tests): rewire u.process consumers to the fail-loud PR #504 contract

[marlon-costa-dc]

Summary

• flext-core PR #504 (merged 8de52fa6b) removed the on_error parameter from u.process (FlextUtilitiesCollection.process). The first processor failure now returns r.fail carrying the processor's original exception, with no skip/collect mode.
• tests/unit/test_utilities_cov.py:37 dropped the removed on_error="fail" kwarg (no behavior change — "fail" was already the only real mode).
• tests/unit/test_utilities_cov.py:43 (the on_error="skip" test) is deleted — that behaviour no longer exists.
• Added test_process_fails_loud_with_original_exception_and_stops_at_first_failure, a behavioural test through the public u facade proving the run stops at the first failing item (later items are never visited) and result.exception is the processor's original raised exception.
• src/flext_cli/_utilities/validation.py and tests/unit/test_cmd_runtime_validation_branch_cov.py were checked for leftover process_mapping modes from flext-cli PR fix(cli): fail loud at the remaining CLI_SAFE_EXCEPTIONS sites; timing-free process tests #203 — none found; that cleanup already landed on the tip.
• make upg re-resolved the lock, pinning flext-core to 8de52fa6b (the exact PR #504 merge commit); make gen regenerated the managed projections that follow from that lock bump (docs, Makefile, pyproject.toml, mkdocs.yml, a couple of generated catalog files).

Evidence (worktree /home/marlonsc/flext-work/v8-process-consumers/flext-cli)

• make upg: exit 0 — lock pins flext-core @ git+...@0.12.0-dev#8de52fa6b117ef54a840ac655c4518aff26ba027.
• make setup: exit 0.
• make gen x2: exit 0 both times; identical working-tree diff after the second run (fixed point).
• make fix: exit 0, 0 errors.
• make fmt: exit 0, 1 file reformatted (unrelated cosmetic reflow in tests/unit/test_runtime_process_descendants.py).
• make check: exit 0, Total: 1 Success: 1 Failed: 0 Skipped: 0, 15 gates at 0 errors (duplication/codemod/boundary/namespace/runtime-census gates suspended by operator authority flext-itpd1.3).
• make test: exit 0, 1193 passed — includes test_process_fails_when_processor_raises, test_process_fails_loud_with_original_exception_and_stops_at_first_failure, test_process_predicate_excludes_items_before_processing, all passing.
• Runtime proof, direct u.process call through the installed flext-cli package (not just pytest):

  success: [10, 5, 2]
  failure ok? True exception= ZeroDivisionError('integer division or modulo by zero') error= Processing failed for item 0: integer division or modulo by zero
  

Test plan

• make check green (15 gates, 0 errors)
• make test green (1193 passed)
• Runtime proof of u.process fail-loud contract through the flext-cli public u facade

🤖 Generated with Claude Code

---

Summary by cubic

Rewires u.process test consumers to the fail-loud contract from flext-core PR #504, which removed the on_error parameter. The first processor failure now returns r.fail carrying the processor's original exception, with no skip/collect mode.

• Dropped the removed on_error="fail" kwarg — no behavior change, since "fail" was the only real mode.
• Deleted the on_error="skip" test, as that mode no longer exists.
• Added a behavioral test asserting the run stops at the first failure, never visiting later items, and that result.exception is the processor's original exception.
• Pinned flext-core in uv.lock to the PR #504 merge commit and regenerated the managed projections (docs, Makefile, pyproject.toml, mkdocs.yml).

^{Written for commit 62b3156. Summary will update on new commits.}

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud