mlem deploy create should re-deploy if model changed

[aguschin]

Saving a model from #235, then deploying it to heroku makes error (as numpy is not installed in heroku container):

$ mlem deploy status myservice
crashed

If you add some lines to dog-bark-translator.mlem:

requirements:
- module: emoji
  version: 1.7.0
- module: numpy
  version: 1.22.3

and try to redeploy:

$ mlem deploy create myservice
⏳️ Loading deployment from myservice.mlem
🔗 Loading link to file://staging.mlem
🔗 Loading link to file://dog-bark-translator.mlem
💾 Updating deployment at myservice.mlem
✅  Service mlem-quick-start is up. You can check it out at https://mlem-quick-start.herokuapp.com/

Nothing changes, e.g. https://mlem-quick-start.herokuapp.com/ still doesn't work and as I can see from heroku logs, model wasn't redeployed.

It could be default behavior to redeploy if something changed (I would prefer this).
It could also be --force option, but I don't see it in mlem deploy create myservice -h.
Even if we use default to redeploy, --force still may be useful.