fix(selfhost): backup-miner.sh / restore-miner.sh default to the pre-rename ~/.config/gittensory-miner state dir

[JSONbored]

Context

scripts/backup-miner.sh and scripts/restore-miner.sh (the DR pair for the AMS miner's local SQLite
stores, #4872) both default their working directory to the pre-rename config path:

# scripts/backup-miner.sh line 20
STATE_DIR="${LOOPOVER_MINER_CONFIG_DIR:-$HOME/.config/gittensory-miner}"

# scripts/restore-miner.sh line 18
STATE_DIR="${LOOPOVER_MINER_CONFIG_DIR:-$HOME/.config/gittensory-miner}"

But the miner's actual, current default config directory — the one every real store path resolver in the
package uses — is ~/.config/loopover-miner, per packages/loopover-miner/lib/local-store.js's
resolveLocalStoreDbPath:

// packages/loopover-miner/lib/local-store.js
const configHome = typeof env.XDG_CONFIG_HOME === "string" && env.XDG_CONFIG_HOME.trim()
  ? env.XDG_CONFIG_HOME.trim()
  : join(homedir(), ".config");
return join(configHome, "loopover-miner", defaultDbFileName);

This resolver is shared by every real store (run-state.js, claim-ledger.js, portfolio-queue.js,
event-ledger.js, etc.) via local-store.js. The rename is confirmed intentional and final in
packages/loopover-miner/CHANGELOG.md: "the miner's default config directory... changed; no
dual-read/alias, per the epic's full-cutover mandate (#5705).
A self-hoster's existing ~/.config/gittensory-miner state does not migrate automatically."

Practical effect: an operator who never sets LOOPOVER_MINER_CONFIG_DIR explicitly (the common case —
the miner itself defaults correctly without it) runs sh scripts/backup-miner.sh expecting it to back up
their real state at ~/.config/loopover-miner. Instead it looks at ~/.config/gittensory-miner, which for
a fresh post-rename install doesn't exist at all, so the script exits 1 with "state dir not found... nothing
to back up" — not backing up the real data unless the operator notices the stderr line (easy to miss in an
unattended cron/systemd timer, exactly the deployment mode scripts/backup-miner.sh's own header comment
targets). restore-miner.sh has the identical wrong default, so a DR restore attempted without an explicit
override also targets the wrong directory and effectively does nothing to the live state.

Requirements

• Change both scripts' default STATE_DIR fallback from $HOME/.config/gittensory-miner to
  $HOME/.config/loopover-miner, matching local-store.js's real default exactly (these scripts already
  only fall back to $HOME/.config, so no other behavior change is needed beyond the directory name).
• Do not add a dual-read/fallback to the old gittensory-miner path — the CHANGELOG is explicit that this
  rename is a hard cutover with no migration shim; these scripts should match that, not reintroduce one.
• Update any comment text in both files still referencing gittensory-miner state where it describes the
  default path (leave historical/issue-number references like #4872 alone).

Deliverables

• scripts/backup-miner.sh: default STATE_DIR fixed to $HOME/.config/loopover-miner.
• scripts/restore-miner.sh: default STATE_DIR fixed to $HOME/.config/loopover-miner.
• A regression test (e.g. under test/unit/) that greps both scripts and asserts the default fallback
  string is loopover-miner, not gittensory-miner, so this can't silently regress again.

Test Coverage Requirements

These are scripts/** shell files, outside src/** and outside Codecov's coverage.include — Codecov
patch coverage does not gate this change directly. Add a real Vitest test (following the existing pattern in
test/unit/miner-docker-compose.test.ts, which reads a config file with readFileSync and asserts its
content) that reads both scripts' source and asserts the correct default directory string appears and the
stale one does not. This gives the fix a real regression check even though Codecov won't score it.

Expected Outcome

Running sh scripts/backup-miner.sh or sh scripts/restore-miner.sh with no LOOPOVER_MINER_CONFIG_DIR
override operates against the miner's real, current state directory (~/.config/loopover-miner), matching
every other part of the package. A fresh post-rename install's scheduled backups actually back up real data
instead of failing to find a directory that no longer exists.

Links & Resources

• packages/loopover-miner/lib/local-store.js — canonical default-path resolver every real store uses.
• packages/loopover-miner/CHANGELOG.md — confirms the hard-cutover rename with no dual-read/alias.
• packages/loopover-miner/Dockerfile — sets LOOPOVER_MINER_CONFIG_DIR=/data/miner explicitly for fleet
  mode, which is why this bug only affects laptop-mode (no-Docker) installs relying on the implicit default.
• Related but distinct from chore(selfhost): docker-compose.yml PROFILES header is missing ams-observability and backup #5812/docs(selfhost): .env.example is missing ~15 vars docker-compose.yml actually reads (backup, AMS exporter paths, OTEL metrics, retention) #5814/fix(observability): LoopoverOrbExportErrorRateHigh runbook references a metric that doesn't exist #5816/fix(selfhost): Terraform firewall never opens Grafana's port for --profile observability #5818/fix(observability): resource-hub.json is missing a link to the Miner usage (AMS) dashboard #5820 (filed earlier today) — this is a different file
  class (shell scripts) with a different failure mode (wrong default directory, not a doc/profile/metric gap).

[linear-code]

JSO-4067