naming users and datasets needs to be unified and enforced

[dustmop]

The recently added dsref package should be the central location for parsing usernames and dataset names. In both cases we should only allow names that start with a lower case letter, then follow with one or more of (lower case letter, number, underscore, dash).

We also need to enforce this naming scheme everywhere. Here are some places I found that are behaving incorrectly:

• [x] running qri setup doesn't validate that the provided username is valid
• [ ] logbook doesn't validate that the username for author init is valid
• [ ] modifying the profile.peername field in config.yaml can create invalid usernames
  • afterwards, running qri save will successfully create a dataset with an invalid username, but logbook will return the error "log: not found"
• [x] qri save --body my-body.csv incorrectly turns "-" into "_" because package varName doesn't match dsrerf
• [x] qri checkout also creates a directory name from the dataset name using varName
• [ ] dataset/validate.go uses ValidName which doesn't allow "-", but this appears to be unused
• [ ] peer messages received over p2p should also be validating received names

Desktop also needs to have parity with these checks

[dustmop]

Reversing the fix for this issue: https://github.com/qri-io/qri/issues/1116.

Related: https://github.com/qri-io/qri/issues/1153 and https://github.com/qri-io/qri/issues/1132

[dustmop]

Related issue that can be fixed at the same time. It's not possible to infer a name using a file that contains unicode:

> qri save --body=здравствуйте.csv 
invalid name: dataset name must start with a letter, and only contain letters, numbers, and underscore. Maximum length is 144 characters