docs(calls): clarify vapifault-worker-not-available vs status.vapi.ai and billing

[Mehayla]

What changed

In fern/calls/troubleshoot-call-errors.mdx, under Phone never rang → Infrastructure errors:

• Expanded the Fix guidance for call.in-progress.error-vapifault-worker-not-available to note that this is a brief, transient capacity blip that can affect a region or a subset of orgs without crossing the threshold to be posted as a status.vapi.ai incident.
• Added a <Note> directly after the Infrastructure errors accordion covering: what to do if vapifault-prefixed errors recur with no matching status page incident (retry, then contact support with call_ids and timestamps if it persists), and a billing clarification (these errors are not billed by default; contact support with the call_id for a credit if you were charged).

No other content, wording, or structure in the file was changed.

[Mehayla]

Why this change

Evaluated the last 7 days (Aug 17–24, 2026) of customer feedback across support tickets, Ask AI chat, Discord, and NPS survey sources, benchmarked against the prior 7 days and a trailing 4-week average.

Most topics were down week-over-week (a broad pattern this week), but Uptime & Availability was the one topic that rose above trend: 83 conversations this week vs. 70 the prior week and a ~74/week 4-week average — a sustained, not one-off, signal.

Drilling into that topic's tickets from this week, at least 5 distinct customers independently reported call.in-progress.error-vapifault-worker-not-available failures during the same Aug 22, 13:00–14:30 UTC window (inbound SIP/Telnyx/BYO-number calls failing immediately with empty logs), and several explicitly said they checked status.vapi.ai and saw no incident, which is why they doubted the "transient capacity issue" explanation:

"getting a lot of these whats the situation i dont see any downtime on VAPI status but it seems there is..."
— customer, support ticket

"This strongly suggests a transient Vapi infrastructure or media-bridge problem rather than a persistent configuration issue. Could you please: 1. Investigate the six call IDs and provide a root-cause analysis. 2. Confirm whether there was an infrastructure, worker-capacity, or media-routing incident."
— customer, support ticket (also asked about billing/reimbursement for the failed calls)

Content check: fern/calls/troubleshoot-call-errors.mdx already documents vapifault-worker-not-available ("Retry. This is a transient capacity issue."), and fern/calls/call-ended-reason.mdx covers the error prefix classification. Neither page explains that brief capacity blips can occur without ever being posted as a status.vapi.ai incident, and neither addresses the billing/credit question these customers were raising in the same tickets.

Classification: Incomplete content — the error code is documented, but the specific scenario customers are hitting (no status-page incident + billing confusion) wasn't covered.

This PR only touches the Infrastructure errors accordion's Fix cell for that one error code and adds one <Note> closing that gap. No other wording, structure, or unrelated section was touched.