L1 RPC client to inject FRBs

[shriharshtendulkar]

Based on discussion at McGill on 19th Sept 2018.

Format for sparse data packet:

int version
uint64 fpga_count_sample_0
uint32 first_sample[16384]
uint32 nsamples[16384]
unint32 total_nsamples
float32 data[total_nsamples]

We'll have a separate python pipeline to make the data for all L1 beams and it will send the injection data to each L1 server well in advance.

[dstndstn]

This is now implemented and ready for review (by Kendrick)

kmsmith137/simpulse#4
#21
kmsmith137/rf_pipelines#26
kmsmith137/ch_frb_l1#27

I implemented a very simple python pulse-injector in rpc_client.py, using simpulse.

One thing that will need to be done is to get the scalings of the frequency channels; currently the units of the RPC are uncalibrated intensity units. When doing injections, I think we will want to call an RPC to retrieve a current estimate of the variance of each channel, and then scale the pulse appropriately, so that it comes out with the correct S/N.

This plot shows some testing I was doing, injecting different "RFI" signals and a pulse.