HCMI sample file is missing important metadata and does not cover the entire repository

[sgosline]

We need to update the current manner in which we capture HCMI metadata to fully track all the samples. Currently there is no tie back tot he case_ids, which are the primary identifiers, nor is there any tracking of cancer diagnosis, tumor vs. normal, or primary vs met. This can be rectified by updating the data capture algorithm using a well-thought-out schema. Using this updated approach we can capture all the data in the current HCMI sample schema:

We should map the following CoderData fields to the hcmi fields. Like the broad_sanger data, there will be multiple rows for a single sampleID.
cancer_type: This should be the 'Clinical tumor diagnosis'
common_name: this should be the 'sample_submitter_id'
other_id: there should be MULTIPLE of these per sample, which can be duplicated

1. case_submitter_id (other_id_source should be 'case_submitter_id')
2. if available: diagnosis_id (other_id_source should be 'diagnosis_id') (there can still be multiple aliquots per dx)
3. if available: treatment_id (other_id_source should be 'treatment_id')
4. sample_uuid (other_id_source should be 'sample_id'). <----this should be 1:1 with improve_sample_id
5. case_uuid (other_id_source should be 'case_id')
6. other_name: add 'tissue_type' here
7. other_name: add 'tumor_descriptor' here

If fixed this should address the need for #185