IRQ Improvements

[1Revenger1]

I'm not entirely sure what to name this PR honestly, since it's all a bunch of small changes. I was helping out @theroadw with a touchpad that keeps the IRQ line asserted and wrote these changes in an attempt to figure help troubleshoot it.

Notes on each of the commits:

• Remove unused isCommunityActive
• Only set IRQ type on resume/register
  • IOInterruptEventSource calls disableInterrupt/enableInterrupt automatically on a level interrupt, so setting the pin type was being called every time a touchpad received an interrupt.
  • VoodooGPIO should save off all the pad configuration registers on sleep so I don't believe this needs to be called on wakeup.
• Always clear interrupt status on IRQ
  • I think 99% of I2C devices uses level interrupts so this was never an issue, but the linux driver does always clear this bit. Datasheets also suggest that this isn't specific to level interrupts.
• Mux pin into GPIO controller logic
  • Copied from the linux implementation, makes sure that the pin wasn't routed to another block of hardware.
• Fix pmode being removed twice from pad config
  • I meant to split this into two commits, as it hides a pretty big change to how interrupts are handled.
  • This leaves interrupts always enabled so that the IOInterruptEventSource disabling an interrupt temporarily doesn't prevent other devices from receiving IRQs. Disabling still masks the interrupt so I think this should be fine.
• Return interrupt type for GPIO pin rather than overall GPIO controller
  • IOInterruptEventSource does test the interrupt type to check if it should be disabling/enabling interrupts. Was a pretty quick thing to add but since everything uses level interrupts, the behavior doesn't really change in the end.

[kprinssu]

Thanks for cleaning up some of the tech debt. This LGTM.