Now works even when the CPU is stuck in a watchdog reset loop, as it would be when the flash is totally blank.

CSW at this point isn't valid if the error wasn't a timeout but it was something else instead, like a FAULT.

Working on making this reliably halt a chip that's stuck in a watchdog reset loop :(

That was a silly library, full of middleware and memory leaks. Debugging libswd was a good way to get familiar enough with SWD to write a replacement really quickly :P

This still doesn't work reliably, it seems to have the same unsynchronization bug that the old code was having. But it's much smaller and easier to understand now!

Some silly bugs are undoubtedly lurking here, but sometimes this will successfully read and write target RAM :)

For the vendor code to make it through string conversions unharmed, we should make it part of the printable ASCII subset. Fixes compatibility with xusb when reading WCID descriptor.

Just saw Paul's note about USB DMA from Flash being sketchy. In the boot loader it doesn't matter since we exec from RAM anyway, but we might as well un-const the WCID descriptor for consistency.

Thought this would make the win32 dfu-util less spammy. No such luck, but it seems like a good idea anyway.

FC-Boot works now, no need for this.