Copy head (r256279) to stable/10 as part of the 10.0-RELEASE cycle.

Approved by: re (implicit)
Sponsored by: The FreeBSD Foundation

Support sending ATH_ALQ messages with no payload.

Extend the timestamp to be a timeval, rather than ticks.

This makes it easier to see TX and RX buffer latencies.

ath(4) ALQ logging improvements.

* Add a new method which allows the driver to push the MAC/phy/hal info
into the logging stream.
* Add a new ALQ logging entry which logs the mac/phy/hal information.
* Modify the ALQ startup path to log the MAC/phy/hal information
so the decoder knows which HAL/chip is generating this information.
* Convert the header and mac/phy/hal information to use be32, rather than
host order. I'd like to make this stuff endian-agnostic so I can
decode MIPS generated logs on a PC.

This requires some further driver modifications to correctly log the
right initial chip information.

Also - although noone bar me is currently using this, I've shifted the
debug bitmask around a bit. Consider yourself warned!

Fix a very incorrect description.

Add my initial cut at driver-layer ALQ support.

I'm using this to debug EDMA TX and RX descriptors and it's really helpful
to have a non-printf() way to decode frames.

I won't link this into the build until I've tidied it up a little more.

This will eventually be behind ATH_DEBUG_ALQ.