fix(mpg): clarify restore command help text

[tripledoublev]

Change Summary

What and Why:

fly mpg restore provisions a new, separately-billed cluster from the backup and leaves the source cluster untouched. The help text didn't say that, and gave no pointer to where --backup-id values come from. A customer reported the command as confusing.

How:

Update the command's short/long help text to state that restore creates a new cluster (billed separately, same organization, provisioned asynchronously) rather than restoring in place, and to point at fly mpg backup list for backup IDs. Help text only — no behavior change, --backup-id description left as-is.

Verified by:

• go build, gofmt, go vet, go test ./internal/command/mpg/... — all clean;
• inspecting the rendered fly mpg restore --help from a built binary;
• confirming each claim against both the v1 and v2 code paths, since mpg restore dispatches on cluster version — the wording is deliberately version-agnostic because the two differ (e.g. generated-name format is <source> restored <timestamp> on v1 vs <source> restored <backup_id> on v2).

Open question for reviewers: this is now on the long side for flyctl help text (~70 words vs a ~10-word median). Happy to trim if it reads as too much — the facts were added in response to the reported confusion, but which of them earn their place is a fair thing to push back on.

Related to:

• superfly/mpg#285

Follow-ups tracked in that issue and intentionally out of scope here: --name to set the restored cluster's name, and --pitr-time for point-in-time restore.

---

Documentation

• Fresh Produce
• In superfly/docs, or asked for help from docs team
• n/a

[miguel-fly]

The long description looks a bit verbose, since all the lines are concatenated without line breaks:

--
Consider reducing the description a bit. For example:

[tripledoublev]

Trimmed the help text in e163911a0 to match the suggested wording; verified the rendered output, and go test ./internal/command/mpg/... passes.