Document t3.storage.dev endpoint, Tigris CLI, and zero-downtime migration

[katieschilling]

Summary

Updates tigris/index.html.markerb to document the canonical Tigris S3 endpoint (https://t3.storage.dev), the Tigris CLI / SDK ecosystem available alongside flyctl, and the zero-downtime migration path. The goal is to give users (and search/LLM tools) a clear, accurate picture of how Fly-provisioned Tigris buckets relate to native Tigris tooling.

What changed

• Lead paragraph now describes Tigris as a globally distributed, multi-cloud object storage service that is natively available as Fly's object storage provider — replacing the older "built on Fly.io infrastructure" framing.
• Second paragraph clarifies cross-region replication (vs. caching) and notes the CDN-replacement use case.
• New Access keys subsection: explains that fly storage create prints the key/secret once, that flyctl can't retrieve them later, and that the dashboard / Tigris CLI are the recovery paths.
• Reworked The Tigris web console subsection: documents both ways into the console (the Tigris Object Storage tab in the Fly Dashboard, and flyctl storage dashboard <bucket>), and the two Tigris sign-in options (Fly.io vs direct accounts).
• New Connect to your bucket section with subsections:
  • Endpoint — https://t3.storage.dev is canonical and recommended for new code; explains that fly storage create currently sets AWS_ENDPOINT_URL_S3=https://fly.storage.tigris.dev (still works) and shows how to switch.
  • Tigris CLI — install via npm install -g @tigrisdata/cli, then tigris login → OAuth2 → Fly.io.
  • Tigris SDKs — links to Tigris's language-specific guides.
  • Snapshots and forks — Tigris-only features that aren't exposed through flyctl; must be enabled at bucket creation; setup uses the Tigris CLI.
• flyctl storage list and fly storage destroy now show Tigris CLI equivalents alongside the existing flyctl commands.
• Migration section retitled and rewritten as Migrate from another S3-compatible provider with zero downtime:
  • Leads with the zero-downtime, lazy-migration framing
  • Defines "shadow bucket" inline since the term is unfamiliar to most readers
  • New How shadow bucket migration works subsection covers reads, default writes, write-through writes, deletes, and listing semantics
  • New Write-through mode: zero-downtime cutover subsection explains why dual-writes make migration reversible
  • Links to the Tigris migration guide for provider-specific setup
• Pricing and Billing now states that Tigris buckets created through a Fly account are billed via the regular Fly bill — no separate Tigris invoice.

Test plan

• Visual review of the rendered page
• Verify links resolve (Tigris CLI docs, Tigris migration guide, console.storage.dev)
• Verify anchor links inside the page (#tigris-cli, #connect-to-your-bucket)

[swilson-fly]

Failing linter checks are because tigris is lower cased in the partials import of other files. This is correct/normal not an issue. Merging.