recover

How to recover from an error: the error's data.recover. Unless it is none, the error's data.next holds one concrete action that does it: an operation with its params, the same request sent again, or an lvcs command line.

Value	Meaning
`wait`	Wait for something outside the caller to change (an attestation, a sync, a holder releasing a range), then try again.
`retry_after`	Send the same request again after the time the error gives, in `data.retry_after_ms`.
`rebase`	Read the document again and redo the change from what it holds now.
`ask`	Ask a participant who can allow it: the owner, or the issuer of the grant the request needs; for a peer whose release is older than this daemon's, its owner to upgrade.
`yield`	Leave the range to the participant working on it, and work elsewhere.
`confirm`	Confirm the request is what was meant, and it goes ahead.
`renew`	Get the grant renewed, or a new grant, from its issuer.
`fork`	Work in a forked workspace of your own instead, and merge it back later.
`join`	Join the project as a participant first.
`fix_request`	Change the request itself: its parameters, its path, or its operation.
`upgrade`	An LVCS-owned file under `.lvcs/` has a newer format than this `lvcs` reads, or a peer speaks a newer wire major than this daemon: install a newer `lvcs`, then run the command again.
`migrate`	An LVCS-owned file under `.lvcs/` has an older format, or none (written before format headers): run `lvcs migrate` in the project root with no daemon running, then run the command again.
`none`	Nothing the caller can do changes the answer.

Cardinality: 13.