Add timestamp to header

[thomascobb]

Currently PCAP has relative timestamps for things like TS_CAPTURE. MAX IV would like to relate this to an absolute timestamp that is provided in the header of the capture. This can be provided by NTP now, but in the future may be provided by the timestamp from the EVR.

[Araneidae]

I like the approach suggested by @valerixb, but this leaves open the question of when capture actually started. Will the ARM timestamp be enough to solve this, or are we also going to want the driver to capture a timestamp when capture begins? This may be tricky, as data only starts to arrive after a buffer's worth has been seen, or a timeout.

[thomascobb]

How about this, for now we add pcap_arm_time which is accurate to NTP (a couple of ms?). In the future we can add a second pcap_enable_time which is raised as an interrupt when PCAP.ENABLE goes high after an ARM, which is accurate to the latency of getting an interrupt into the ARM (not sure how accurate this is?)

[Araneidae]

Sounds good to me. We'd have to add an ENABLE interrupt, or capture a hardware timestamp in firmware.

[coretl]

PandABlocks/PandABlocks-FPGA#88 needs to be done first