Read attributes inaccessible via multiprocessing pool

[dkoslicki]

For the Read class, it appears that the sequence attribute is inaccessible when passed to a parallel pool. As a MWE, consider:

import multiprocessing
import khmer

def my_func(record):
	return len(record.sequence)

parser = khmer.ReadParser('test_file.fq')

# read in a few records
read_list = []
i = 0
for record in parser.iter_reads():
	read_list.append(record)
	if i >= 10:
		break
	i += 1

pool = multiprocessing.Pool(processes=2)

res = map(my_func, read_list)  # works
res2 = pool.map(my_func, read_list)  # doesn't work: AttributeError: 'Read' object has no attribute 'sequence'.

print(res)
print(res2)
pool.close()

This results in:

Traceback (most recent call last):
  File "MinimumWorkingError.py", line 22, in <module>
    res2 = pool.map(my_func, read_list)  # doesn't work: AttributeError: 'Read' object has no attribute 'sequence'.
  File "/usr/lib/python2.7/multiprocessing/pool.py", line 251, in map
    return self.map_async(func, iterable, chunksize).get()
  File "/usr/lib/python2.7/multiprocessing/pool.py", line 567, in get
    raise self._value
AttributeError: 'Read' object has no attribute 'sequence'.

Same thing happens in python2 or python3 (khmer.__version__ 2.1.1). Unsure, but #645 may be relevant.

Motivation is I'd like to stream in a sequence file and immediately pass it off to pool workers for downstream analysis.