sync tmpos with error handling

[JrtPec]

This pull requires some updates to TMPO first: flukso/tmpo-py#17 and flukso/tmpo-py#19.

Long story short:

• While syncing tmpo, I sometimes encountered a JSONDecodeError. This turned out to be a HTTP-error 500 (Flukso server side error) in disguise. I suspect a faulty TMPO-block or something. flukso/tmpo-py#17 raises the HTTP-error instead of the JSONDecodeError.
• An error like this can screw up your entire script, while all other sensors are probably fine. This pull requests adds the option to ignore single sensors that raise HTTP-errors while syncing. hp.sync_tmpos() has an argument http_errors, which is 'warn' by default, so the error is transformed into a warning. If you make it 'ignore' the error is ignored completely. If it is 'raise' (or anything else really), the error is raised.
• If an error is raised during a tmpo database connection, the connection ends abruptly and you'll get 'database is locked'-errors when you try again. flukso/tmpo-py#19 addresses that issue.

[coveralls]

Coverage decreased (-0.4%) to 70.393% when pulling fdf4c51d89569abd732daef7a4cc0c5c3de89513 on JrtPec:ignore-tmpo-sync-errors into 167982b28476eaceb5e3e809f60333fa4015fed5 on opengridcc:develop.

[saroele]

Well done, I can confirm that this works. I got stuck on the JSONDecodeError tonight and tried your pull request. As you said, I updated the tmpo-py first.

Let's try once more to push the changes to tmpo-py upstream so we avoid creating forks. @icarus75 can you take care of the pull requests, mainly flukso/tmpo-py#17 and flukso/tmpo-py#19.

[icarus75]

@saroele @JrtPec Thanks for the heads-up. PR #15,16,17 & 19 have been merged into tmpo-py master. An updated package (0.2.3) has been pushed to Pypi. Let me know if there are any issues with this latest version.