rqlite-10.0.5-bp160.1.1

- Update to version 10.0.5: significantly improves clustering,
  data integrity guarantees, and makes managing rqlite easier.
  * introduces a major improvement to the Raft Snapshot and Log
    Truncation process. Previously, a Snapshot stream to a Follower
    would block the Leader from taking new Snapshots. Normally a
    subsequent retry of the snapshot operation would succeed, but a
    persistently slow Follower that kept requiring Snapshots could
    starve the Leader entirely. With v10, Snapshotting on the
    Leader is no longer blocked by any other operation happening in
    the Snapshot Store, including streaming a pre-existing snapshot
    to a Follower.
  * Concurrent read and write access to the database has been
    improved. In previous releases Raft snapshotting -- and, as a
    result, writes -- could be blocked for an extended period if
    there was a long-running read. In v10 the Raft snapshotting
    process is canceled and retried later if a long-running read is
    active.
  * v10 also uses CRC32 checksumming more comprehensively to detect
    issues such as file system corruption and inadvertent
    modification of the underlying data files by systems other than
    rqlite.
  * introduces a new built-in console app, making it more convenient
    to manage a rqlite deployment. The console app is available at
    http://localhost:4001/console by default.
  * Upgrading from v7 or later is seamless and has been extensively
    tested, though upgrading first to the latest v9 release is
    recommended. Upgrading to v10 supports rolling upgrades, but you
    cannot join a new v10 node to a v9 (or earlier) cluster. Upgrade
    your existing cluster to v10 before adding new nodes. Back up
    your rqlite system before upgrading.
  * cli change: -on-disk-path has been removed. It provided little
    benefit while making it too easy to corrupt a node. To upgrade a
    rqlite system which uses this flag first backup your rqlite node
    deploy a new v10 system, and then initialize the new deployment
    using the backup.
  * cli change: -raft-timeout is now -raft-heartbeat-timeout, to
    better reflect its purpose.
- includes fix for CVE-2026-33814: golang.org/x/net/http2: infinite
  loop in HTTP/2 transport when given bad SETTINGS_MAX_FRAME_SIZE
  (boo#1265706)

- Update to version 9.4.5:
  * Rely soley on SQLite busy for WAL truncate
  * Prevent deadlock situation during raft snapshots

- Update to version 9.4.3:
  * Remove mustTruncate as it can result in a deadlock
  * Fix: INSERT/UPDATE with RETURNING clause will not abort
    transaction on erro

- Update to version 9.4.1:
  * Add proxy layer between HTTP layer and Store
  * Use maps and slices packages where possible
  * Check if DB context contains an error and return it if so

- Update to version 9.4.0:
  * Add /db/sql API to demonstrate SQL rewriting
  * Upgrade SQL parser
- includes changes from 9.3.21:
  * improvements to SQLite WAL checkpointing.
- includes changes from 9.3.20:
  * Switch to Synchronous=FULL mode when checkpointing database
  * Ensure checkpointing doesn't permanently change Synchronous mode
  * Explicitly fsync WAL file copied to Snapshot store
- includes changes from 9.3.19:
  * various logic improvements

- Update to version 9.3.18:
  * Upgrade SQL parser

- Update to version 9.3.15:
  * Upgrade rqlite-go-sqlite3 and SQL parser bundled dependencies

- Update to version 9.3.14:
  * Increase WAL Checkpoint busy timeout to 250ms
  * Implement a must-checkpoint-WAL after multiple failures policy
  * fix SQL comment handling in CLI

- Update to version 9.3.11:
  * Correctly handle EXPLAIN QUERY PLAN for mutations

- Update to version 9.3.10:
  * Remove any temporary WAL files if persisting a Snapshot fails
    or is not even invoked
  * Improve FSM logging
  * Order alphabetically output of shell command .tables