Possible issue with RMW thread safety in [de]assert_risc_reset

[mcraigheadTT]

Describe the bug
These functions do a RMW of the soft reset register, but if multiple threads call this function concurrently, they could get confused. Note that they used to do a write only until recently, so it's possible that this is a regression.

Stack trace
none

To reproduce
currently none known

Expected behavior
A thread safe implementation (or) higher level invariants that ensure thread safety at this level.