pal_finder: fix 'fastq_subset.py' utility for big input Fastq files

[pjbriggs]

PR which updates the read counting in the fastq_subset.py utility used by the pal_finder tool so it can better deal with large Fastq files.

Without the fix the read counting attempts to read the entire Fastq into memory, resulting in the following error if the Fastq file is too big:

Traceback (most recent call last):
  File "XXXX/toolshed.g2.bx.psu.edu/repos/pjbriggs/pal_finder/52dbe2089d14/pal_finder/fastq_subset.py", line 133, in <module>
    nreads = count_reads(args.fastq_r1)
  File "XXXX/toolshed.g2.bx.psu.edu/repos/pjbriggs/pal_finder/52dbe2089d14/pal_finder/fastq_subset.py", line 70, in count_reads
    buf = fq.read()
MemoryError
FATAL ERROR pal_finder failed to complete successfully

The fix buffers the reading into chunks which should fit into memory and so avoid the error.