misc. bugfixes

[meeg]

• asm_v2.py needed an update to add tProc v2 revision 25 (added in [#333] Fix for tProcV2 doesnt pause when dispatcher FIFOs are full #345) to the list of supported revisions.
• tProc v2 external start (introduced in changes to improve multiboard sync #342) had a bug where if you used external start, but the tProc didn't promptly get a start signal, the software would decide the program had completed already (because the shot counter would have a stale value from the previous round or the previous program), and you'd get stale data. Fixed by resetting the tProc before starting.
• Programs with a large number of triggers in a shot (the most common example is "super-decimated" readout where you read out a lot of short accumulated windows) could end up confusing the streamer code - it would decide that the optimal "stride" (the number of shots to build up in the buffers before reading them out) was 0 shots. This bug may have been introduced in changes to improve multiboard sync #342, though may well be older than that. Fixed by setting a minimum stride of 1 shot.
• The SPI interface code in ipq_pynq_utils has been cleaned up a bit and tweaked so the spidev driver is always reloaded (so it works even if nobody has run xrfclk yet). This is only used for debugging, nobody should be affected by this.