Add NoDma I2C master async bus implementation

[tullom]

Closes #327

Some kicking off questions:

• The reason I need to make a new_async_nodma() fn instead of hijacking new_async() is because new_async() requires a dma_ch object that impls Peripheral and impls an interrupt with MasterDma. I tried creating structs to impl all that but I hit a wall.
• I added documentation saying NOT to use this and to prefer new_async(), except in the cases of the Flexcomm not having DMA support. Is this enough or do you think this should be gated behind some sort of feature flag to reduce the possibility of misuse?

EDIT:
I was able to create a NoDma structure. I had to minimally modify the DMA api, specifically reserve_channel() now returns an option that will return None if NoDma is used.

[tullom]

FYI @jerrysxie @felipebalbi
I was able to create a NoDma structure. I had to minimally modify the DMA api, specifically reserve_channel() now returns an option that will return None if NoDma is used. Needed to slightly modify the UART driver and hashcrypt driver as well (is surface using these APIs? I can go and send out PRs to fix if needed) to return errors if NoDma is used since it doesn't make sense if NoDma is used.

Tested all touched drivers via examples on hardware and it works flawlessly.