ci(fleet): add retries and dryrun example repos to the roster

[joshua-temple]

Problem

Two newly provisioned public example repos, cascade-example-retries and cascade-example-dryrun, exist (token, branch protection, pinned v0.16.5) but are not in the fleet roster, so the fleet never fans out to them.

Fix

Add both to the canonical roster in the plan/select job of fleet-e2e.yaml (the REMAINDER list) and to the in-sync roster in suite-rc-repin.yaml, so they flow through plan -> remainder matrix -> repin like every other light repo. Both are standalone: retries has only an internal dependent deploy (not a cross-repo dependency on primary), so neither belongs in DEPENDENTS.

• cascade-example-retries: deploy retry rescue, and a dependent deploy gated on a retried dependency's retry shims.
• cascade-example-dryrun: dry-run promote and rollback, and the native-deployment dry-run guard that skips real deploys.

The fleet now fans out to 14 example repos.

Docs updated: internals/coverage-matrix.md and internals/testing.md topology tables list both new repos and what each covers.

Verification

• actionlint clean on both modified workflows.
• Config-only change: no cascade generator/manifest/schema surface touched, so no generated output and no Workflow Drift Check impact. fleet-e2e.yaml and suite-rc-repin.yaml are hand-authored maintainer CI, not generated.
• Both new repos flow through the same roster into the remainder matrix and the repin path; no separate stale hardcoded list left behind.