Review err checks from RunCommand invocations

In some situations we have the following pattern:
```go
cmd, err := RunAnnexCommand(...) // or RunGitCommand
if err != nil || cmd.Wait() != nil {
// do error stuff
}
```

In some cases, we may need the error that's returned from `cmd.Wait()`, which we don't hold onto. Also, the error returned from the RunCommand setup function is rarely (if ever) informative, so there's no reason to keep that one around.

Search the repository for RunAnnexCommand and RunGitCommand invocations, then inspect how each command handles the returned setup error and cmd.Wait(). Done means relevant callers preserve useful Wait errors and stop retaining setup errors that are not informative.

Written by the indexing model from the issue text.