Audit auth/login options end-to-end and rewrite the auth/login docs

Goal

Make it easy and obvious for an agent to log in on behalf of the user. The end
state is great agent UX around auth: an agent should be able to figure out, with
confidence, how to authenticate the user for any supported provider without
guessing, and the docs should reflect exactly what the code does.

What to do

Comprehensively audit everything we have built in around authentication and
login, then rewrite the auth/login docs to match reality and serve agents well.
Treat the code as the source of truth and reconcile the docs and tests against
it. Be thorough; the point is to leave nothing undiscovered.

Areas to definitely review and think through

• The full CLI surface for logging in and managing auth (commands, subcommands,
  flags, aliases, conflicts, and how they compose).
• Every provider we support and how login works for each, including credential
  discovery order, where credentials are stored, and relevant env vars.
• The scriptable / non-interactive login paths and their lifecycle, versus the
  flows that require an interactive terminal.
• The credential-import / external-auth-source behavior and where it is gated.
• The in-TUI login experience and whether it matches the CLI.
• All existing auth/login tests, what they cover, and where coverage is thin.
• Anywhere the current docs (e.g. OAUTH.md, README auth sections, telemetry
  notes) have drifted from the code.

For each of the above, think from the perspective of an agent trying to log in
for a user: is the right command discoverable, is it scriptable, what is the
exact recipe, and where would an agent get stuck or have to guess?

Deliverables

• A comprehensive, code-verified inventory of every login/auth option, with
  pointers to where each lives in the code.
• A clear picture of test coverage and gaps.
• Rewritten auth/login docs that are accurate, complete, and optimized for an
  agent (and a user) to quickly know how to log in for any provider, including a
  crisp non-interactive/agent section.
• Call out any UX gaps that make agent login harder than it should be, with
  suggestions for follow-ups.

Starting points

• src/cli/args.rs, src/cli/login.rs, src/cli/login/scriptable.rs
• crates/jcode-app-core/src/external_auth.rs
• crates/jcode-base/src/auth/ (per-provider modules + tests)
• OAUTH.md, README.md, TELEMETRY.md

Task for Fable 5.

Start with src/cli/args.rs, src/cli/login.rs, src/cli/login/scriptable.rs, and crates/jcode-app-core/src/external_auth.rs, then review crates/jcode-base/src/auth/ and its tests. Compare the code and test coverage with OAUTH.md, README.md, and TELEMETRY.md. Done means a code-verified inventory, accurate agent-focused auth/login documentation, coverage gaps, and identified UX follow-ups.