Optimize implementation of the Register Level Access gateway

For pull request https://github.com/ARMmbed/uvisor/pull/233 the turnaround time can be decreased for single-bit writes by using bitband writes. That way the code can return already while executing the read/mask/write on the bus independently instead of waiting for the read and the resulting write to start.

@AlessandroA

Start by reviewing pull request #233 and locating the Register Level Access gateway implementation for single-bit writes. Investigate whether bit-band writes can let the gateway return before the bus read/mask/write completes; done means reduced turnaround time without changing write behavior.

Written by the indexing model from the issue text.