bench(lint): baseline — no prefer-direct-reexport rule

[mainframev]

Purpose

Baseline for measuring the lint performance cost of the prefer-direct-reexport ESLint rule.

This branch is plain bench/base (upstream master) plus the benchmark CI job only — it contains no rule code.

Compare against: bench/lint-with-rule (the "with rule" PR), which is this branch plus the 4 rule files and nothing else.

Method

Both PRs run a byte-identical lint-benchmark job (literally the same commit), so the harness is not a variable:

yarn nx run-many -t lint --all --parallel=4 --skipNxCache --excludeTaskDependencies

Flag	Why
--all	identical project set on both sides; nx affected would skew asymmetrically
--skipNxCache	lint is cache: true; otherwise we would time cache replays
--excludeTaskDependencies	nx.json sets lint.dependsOn: ["build"]; keeps build time out of the measurement
no --nxBail	the with-rule side has known violations; bailing would abort early and under-report time

Run 3× per PR; the fastest run is the least noise-contaminated estimate.

A second, runner-independent measurement runs TIMING=25 eslint src on react-components/react-button/library and dumps ESLint's per-rule profiler table into the job summary. This matters because the config uses projectService: true (type-aware linting), so TypeScript program construction may dominate total lint time and swamp the rule's own cost in wall-clock terms.

Fork CI enablement

Existing jobs are left untouched — they are gated on github.repository_owner == 'microsoft' and so skip automatically in this fork, leaving the runner to the new job. The new job is gated on github.repository_owner == 'mainframev'.

Throwaway branch; not intended to merge.

[mainframev]

Baseline side of the lint benchmark. Full results and analysis: #34 — summary: the rule adds +1s to a 1148s full workspace lint (+0.09%), below the runner noise floor, and profiles at 0.396 ms per package via ESLint's own profiler.