Restoration workers aren't restarted with the wallet

[KtorZ]

Release	Operating System	Cause
next	Windows & OSX & Linux)	Code

Context

When creating a wallet via the API, we also start a restoration worker in charge of keeping the wallet in sync with the chain. There's one dedicated worker per wallet, and they're fairly resilient to network issues.
However, when restarting a wallet using an existing database that already contains a wallet, not restoration worker is started. Though, the wallet exist in DB and can't be recreated for instance.

Steps to Reproduce

1. $ cardano-wallet launch --state-dir .state
2. $ cardano-wallet wallet create ...
3. SIGINT or CTRL-C on the launcher
4. $ cardano-wallet launch --state-dir .state

Expected behavior

We should see log activity for a restoration worker keeping the wallet in sync.

Actual behavior

Nothing happen, the wallet exists in the DB but no restoration worker in background.

---

Resolution Plan

• Upon starting, lookup existing wallets.
• Start a worker for each wallet.

PR

Number	Base
#486	master

QA

• Add an integration tests which does the following steps (see LauncherSpec)

  1. Start a wallet server (on an empty state, with persistence)
  2. Create a new wallet
  3. Immediately kill the wallet server
  4. Re-start a wallet server, using the same database file
  5. Expect the previous wallet to exist and eventually be restored.

• Note that the test is only a best-effort and will sometimes give false positive (but shouldn't give any false negative!). This is because we are running the test on a fresh new blockchain so, depending on when the restoration kicks in, it might have time to restore the wallet before we even kill the server (:rocket:). This happens fairly rarely in practice, and shouldn't give any false alarm anyway.

[piotr-iohk]

lgtm