VDIF Handling Updates

[cplee1]

A few updates to the VDIF handing and writing. The code has been compiled and tested on Garrawarla, and the output has been tested using DSPSR and CDMT/prepfold. The bug fix does not appear to fix the issues with pulse broadening at short time scales (#45), but it does fix and issue I was having in CDMT where there were artefacts appearing every 1 second (after passing through the convolving filterbank).

Updates from @robotopia:

• 7522697: Adding back in the -n option to the beamformer to chunk the processing.
• f0433e7: Fixed bug where the first second of VDIF data was all zeros. This addresses Investigate/Fix empty output for first second of VDIF data #26.

Updates from @cplee1:

• 99f8c2a/4242a19: Updated the ASCII header file which gets written for each VDIF file. Now includes NPOL (a new requirement in DSPSR) and some of the VDIF packet header info (for convenience/debugging and human readability). This addresses Improve VDIF ASCII header text to better represent the output #47.
• ca32e97: Flattened the multi-dimensional array detected_beam containing the fine-channelised beamformed voltages, and renamed it to data_buffer_fine so that it's not confused with a buffer containing detected output.

Example 1: Below is the beginning of the detected timeseries at a time resolution of 12.5 us, after dedispersing to zero DM. Instead of 1 full second of zeros, there is only ~1.5 ms of invalid data. This is where the PFB hits the edge of the data.

Example 2: Below is a coherently dedispersed detection of J0437-4715, from cdmt/prepfold and from dspsr/pdmp. The profile still looks smeared - even more smeared than the incoherently dedispersed profile in Bhat et al. 2018.

Example 3: The new ASCII header file format.

HDR_VERSION 0.2            # Version of this ASCII header
MWA_CAPTURE_VERSION 0.1    # Version of the Data Acquisition Software
MWA_SAMPLE_VERSION 0.1     # Version of the FFD FPGA Software
VCSBEAM_VERSION v4.2-22-gd9cb093

TELESCOPE    MWA                    # telescope name
MODE         PSR                    # observing mode
INSTRUMENT   VDIF                   # instrument name
DATAFILE     G0057_1253991112_04:37:15.90_-47:15:09.11_ch109.vdif   # raw data file name

UTC_START    2019-10-01T18:51:34    # yyyy-mm-dd-hh:mm:ss
MJD_START    58757.785810           # MJD equivalent to the start UTC

SEC_OFFSET   3000.000000            # seconds offset from the start MJD/UTC
MJD_EPOCH    58757.820532           # MJD of the data epoch

SOURCE       unset                  # name of the astronomical source
RA           04:37:15.90            # Right Ascension of the source
DEC          -47:15:09.11           # Declination of the source

FREQ         139.520000             # centre frequency on sky in MHz
BW           1.280000               # bandwidth in MHz (-ve lower sb)
TSAMP        0.781250               # sampling interval in microseconds

NBIT         8                      # number of bits per sample
NDIM         2                      # dimension of samples (2=complex, 1=real)
NPOL         2                      # number of polarisations observed

[bwmeyers]

Okay, this looks good to me. I see you removed the UTC_START keyword from the VDIF header, as discussed. Assuming all this is tested at some level (which I think it has been based on the plots shown here and offline), will merge it into the main branch ASAP!