indexd.py should coerce types where possible

[znatty22]

Right now if try to submit:

{
  'availability': 'Immediate Download',
 'data_type': 'Aligned Reads',
 'external_id': 's3://kf-seq-data-bcm/chung/HKT73CCXX-5.hgv.bam',
 'file_format': 'bam',
 'file_name': 'HKT73CCXX-5.hgv.bam',
 'hashes': {'etag': '98ebe560095b231a3050812cc58a8b84-8100'},
 'is_harmonized': 'False',
 'reference_genome': 'hs37d5',
 'size': '67939600949',
 'urls': ['s3://kf-seq-data-bcm/chung/HKT73CCXX-5.hgv.bam']
}

Indexd errors because it says size must be an int. I think indexd.py should coerce types where possible (maybe only size?). Dataservice does this for other fields (prob either via sqlalchemy or marshmallow), so the same behavior should be implemented for indexd fields too.