Add MD5 checksum utility

[erikyao]

Datasources in myvariant.io may contain large files to download (e.g. dbsnp release 155 has 380GB). Due to various reasons (like FTP connection issues), the download may be incomplete, leading errors during the uploading processes.

Some of the datasource has MD5 checksum files available. It would be nice to download those *.md5 files as well and validate the data files in the post_dump phrases.

The validation in bash is quite straight forward. Each .md5 file is essentially a tuple of (checksum, filename). md5sum -c will read a .md5 file, re-calculate the checksum for that filename and match it with the origin checksum. E.g.

(venv) myvariant@su09:/data/hub/myvariant_hub/dbsnp/155$ cat refsnp-chr16.json.bz2.md5 
aa8f0ec9c4752ea34dff2ae309d2a239  refsnp-chr16.json.bz2
(venv) myvariant@su09:/data/hub/myvariant_hub/dbsnp/155$ md5sum -c refsnp-chr16.json.bz2.md5
refsnp-chr16.json.bz2: OK

It's also feasible in python with built-in hashlib.md5(). See Generating an MD5 checksum of a file. Performance of feeding file content to hashlib should be taken into account before developing a MD5 helper class/function.

[newgene]

@erikyao Looks like post_download or post_dump might be the good place to add the md5 check for a data src like dbsnp:

https://github.com/biothings/biothings.api/blob/181a36fc2d5f782bb3608ec032891b0eaa9e7e1d/biothings/hub/dataload/dumper.py#L162-L172

[newgene]

see an example here from mychem: biothings/mychem.info@79f704b
and a small change biothings/mychem.info@6fa8937