Pytelemetry

Instantaneous queue size provides too sparse information
Added rolling average on top of it as a better profiling metric

Rx queue size instantaneous and max can now be fetched using `serialtransports.stats()['rx_in_waiting']` and `serialtransports.stats()['rx_in_waiting_max']`

Measurement of various parameters at serial and protocol level
- received bytes amount
- received chunks amount (several bytes read at once)
- Corrupted frames by CRC, header, topic, payload
- etc

These measurements will allow users to detect issues inside their embedded directly from the CLI, such as
- too high throughput
- undersized TX buffer

Caught exceptions in telemetry.transports.serialtransport and logging them

- Added Travis CI integration
- Fixed decoding error/exception on 64 bits systems (Linux and MacOS)
- Fixed readme formatting on Pypi