GCP Filestore NodeLink refused in us-central1

Found by the Dev oracle run (Task 8781, report artifact 8129). Supersedes draft #8879. The epic's "NodeLink may work on GCP" is therefore not yet delivered. Reproduction (observed): app.dev.jaah.ai, New NodeLink › Storage › New GCP Filestore (connection t8781-gcp, us-central1; cost note "Filestore Regional: at least 100 GiB and 2,000 IOPS, about $115 a month") › Provision → failed: "GCP refused the request: capacity less than 1TB is currently not available in this region." Cause: apps/api/app/gcp/filestore.py:43 `TIER = "REGIONAL"`, :46 `CAPACITY_GIB = 100`, :94 `capacityGb` — the 100 GiB Regional minimum is not offered in us-central1. Requirements: 1. Provision a Filestore tier/capacity GCP accepts in the connection's region (e.g. fall back to ZONAL/BASIC_HDD at its minimum, or the region's supported minimum), and make the console cost note state what is actually created. 2. When GCP refuses for capacity, the error says what to change (region or tier), not only GCP's text. Out of scope: Azure Files, AWS EFS, cross-cloud attach, mount wire. Done means: - On dev in us-central1 a new GCP Filestore NodeLink provisions to ready, attaches to two GCP nodes, an account logged in on one is usable on the other (screenshots) — epic AC6.1–AC6.2. - Unit test pins the tier/capacity choice and the refusal message; quick gate green. - The created Filestore is deleted through the console afterwards.

New GCP Filestore falls back to Regional 1 TiB when GCP refuses 100 GiB; capacity refusal names region/tier; cost note states both. Gates green, #6587 merged, dev deploy OK. Live dev oracle not run (console can't delete Filestore → draft #8890). Memory: none applied.