Elastos Node for Ubuntu v1.2.2

Recovery-day release. Includes everything since v1.2.0.

Update with:

node.sh update_script
node.sh version          # expect v1.2.2

Fixed

• A failed chain update reported success. Every *_update function tested chain_prepare_stage with a bare return, which yields the status of the successful test rather than the failure, so a download or extract failure exited 0 while the node kept running its previous binary. All ten call sites now propagate the real exit code, and all_update names the chains that failed.
• Downloads could not detect an HTTP error. curl -O was called without -f, so a 404 or 503 body was written into the archive and curl exited 0. Downloads now use curl -fL with timeouts and retries, and verify a published .sha256 when one is available.
• uninstall destroyed the credentials it backed up. It archived ela/keystore.dat, then deleted the keystore password files and the side chain keystores in the same sweep. It now archives every keystore together with its password file and refuses to delete anything if that archive fails.
• ela restart reported success while refusing to act, so a configuration change applied that way was never live. It now returns a distinct status and prints the stop and start sequence.
• ela consensus applied nothing in either direction (v1.2.1 regression). The handler read the wrong positional argument after the dispatcher shift, so the command printed the current state and exited without changing it.

Added

• ela consensus [status|on|off] sets the DPoS consensus role. The write is atomic and validated, the original is kept as a timestamped backup, and file permissions are preserved. This is the setting inside the ela daemon, not the separate arbiter relay service.
• ela rewound reports whether a node may safely rejoin consensus after a recovery: binary version, store tip, and any leftover height-named checkpoint above the rollback target. Read-only, and it runs against a running node.
• binary prints a one-line build fingerprint per chain, so an operator fleet can confirm every node runs the same build.
• ela update help documents -n, which installs without starting the daemon.