Spend Instagram's logged-in session sparingly, and stop using it once Instagram logs it out

[pwikstrom]

The first run with the cookie fallback scraped 14 gated posts in about a
minute -- two logged-in API calls each, two threads, no pacing -- and
Instagram then logged the session out. It answers a logged-out session's API
call with a 200 redirect to its login page; yt-dlp falls back to logged-out
extraction only when that redirect arrives as an HTTP error, so it surfaced
as "Failed to parse JSON", a retryable unknown, and the scraper sent 87 more
attempts into the dead session before the transient-storm guard stopped it.

• A logged-out session is a new session_expired verdict. It ends every
  further logged-in request for the run (later gated posts skip the cookies),
  keeps the items queued, charges neither retry budget, stays out of both
  storm guards, stops the run, raises a scraper alert asking for a fresh
  login -- ahead of the storm alerts, since a platform keeps one alert -- and
  holds the enrichment supervisor off until a run completes without it.

• Logged-in requests are spaced across threads by
  [misc] scraper_instagram_auth_interval (default 20 s). Public posts keep
  anonymous speed.

• The media leg downloads from the info dict the metadata leg already
  extracted, so a gated post costs one logged-in call instead of two; only a
  retry re-extracts, paced. Both legs now select the same format: an info
  dict processed with yt-dlp's default DASH selection drew HTTP 403 from the
  CDN when a second instance reprocessed it (cookies and impersonation were
  ruled out). Measured live: one extraction per post, where it was two.