Add dump function to write counts to file

[Adamtaranto]

Add KmerCountTable method .dump_hashes() to write sorted hash:count pairs to tab-delimited output file. Can also return as a list of tuples for use in Python.

Added tests.

Addresses issue #24, but does not add an option to output kmer:count pairs at this time.

Example data:

import oxli

# Demo table
kct = oxli.KmerCountTable(ksize=4)
kct.count("AAAA")  # Count 'AAAA'
kct.count("TTTT")  # Count revcomp of 'AAAA'
kct.count("AATT")  # Count 'AATT'
kct.count("GGGG")  # Count 'GGGG'
kct.count("GGGG")  # Count again.

# Hashes
#  17832910516274425539 = AAAA/TTTT
# 382727017318141683 = AATT
# 73459868045630124 = GGGG

Default sort on counts then on keys:

kct.dump_hashes()
>>> [(382727017318141683, 1), (73459868045630124, 2), (17832910516274425539, 2)]

Optional sort on keys:

kct.dump_hashes(sortkeys=True)
>>> [(73459868045630124, 2), (382727017318141683, 1), (17832910516274425539, 2)]

Sorted hash:count pairs can be written to a tab-delimited text file by specifying an output target:

# Write tab-delimited records to kct.dump
kct.dump_hashes(file="kct.dump")

If no output file is specified, records are returned as list of (hash,count) tuples (as above).
This list can be convert to a pandas dataframe:

import pandas as pd
table_dump = kct.dump_hashes()
df = pd.DataFrame(table_dump, columns=['Hash', 'Count'])
print(df)
>>>
  '''
                     Hash  Count
  0    382727017318141683      1
  1     73459868045630124      2
  2  17832910516274425539      2
  '''

If table is empty, returns empty list:

empty_kct = oxli.KmerCountTable(ksize=4)

empty_kct.dump_hashes()
>>> []

[Adamtaranto]

Allow Closes #24

Will revisit .dump() for writing canonical kmers in place of hashes when we have a solution for storing hash:kmer pairs in the KmerCountTable.

[ctb]

curious about needing to sort - for large k-mer collections, this is going to cost real memory. wouldn't a better default be to not sort, and then allow sorting?

[Adamtaranto]

My thinking was that it would be useful to default sort on counts then hashes:

• Gives same dump file irrespective of the order that kmers were consumes (same data, same output)
• Easy to do a quick visual sanity check on the output with head / tail

Opt 1: Add a sortcounts opt (+ logic to deal with both sortkeys and sortcounts being set)
Opt 2: Keep default sortcounts, add opt nosort

Leaning toward Opt 1, can just use sortcounts in the example docs.

[ctb]

if you're up for adding sortcounts option now, that'd be great - but can also be added later. (since I tend to work in bursts, I'll often punt things like this to an issue to be tackled later. goal is to not forget, but not necessarily pressure ourselves to do things quickly :))

[ctb]

nosort is also fine, too!

[Adamtaranto]

• rename func to .dump()
• Add an option kmer=True to output kmers instead of hashes (rather than two separate funcs) later when we have hash:kmer table option.
• Default unsorted output.
• Add sortcounts.
• Error if both sortcount and sortkeys set