wl1271: Improvements to the TX path

- Fix a TX result overflow problem that was present in the TX path and visible
  with at least linksys AP's (probably any AP with high throughput capability.)
- Optimize TX by writing FW trigger for a group of TX frames instead of
  each and every frame.
- Slightly optimize the TX path code.

Signed-off-by: Juuso Oikarinen <juuso.oikarinen@nokia.com>
Reviewed-by: Kalle Valo <kalle.valo@nokia.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>