feat: support repeatable --test-id flag in eval run

[christso]

Summary

agentv eval run --test-id currently accepts a single glob pattern. It would be useful to support repeatable --test-id flags (like --target already does) to filter by multiple specific test IDs in a single run.

Use case

When running evals in CI via GitHub Actions, we expose test_id as a workflow input. Users sometimes want to run a handful of specific tests across suites without writing a glob that matches them all.

Example:

agentv eval run evals/**/*.eval.yaml --test-id "fsa-shipment-mode" --test-id "schema-qualification-*"

Current workaround

Glob patterns can cover some cases (--test-id "fsa-*"), but there's no way to specify multiple unrelated test IDs without running separate commands.

Proposal

Make --test-id repeatable (OR logic — run test if it matches any pattern), consistent with how --target already works.