Improved documentation, and other small changes.

py core:
- add builtin round function
- gc.enable/disable still allows manual GC (as per CPython)
- fix builtin callable for user-defined types
- fix bug for right shift of small int by large amount

unix port:
- add "fast" version
- improve uctypes, including sizeof function

stmhal port:
- support for extended CAN frames
- USB_VCP works with select
- experimental support for RTS/CTS in UART
- make UART bits count number of data bits, not incl parity
- pyb.freq now allows 8MHz and 16MHz

docs:
- revamp and merge all inline docs into Sphinx framework
- add a few more tutorials