Mcl.sickbay

This release simply adds `unknown` as a possible value to the `Mode` enumeration, thus satisfying [issue 21](https://github.com/MCLConsortium/mcl.sickbay/issues/21).

This release increases the limits on sizes in `Biospecimen` for the following fields:

- `specimen_ID_local` 25 → 200
- `specimen_parent_ID` 50 → 200

This release accommodates [listings in the CDE changelog](https://mcl.nci.nih.gov/resources/standards/mcl_cdedictionaries_changelog_v11.xlsx) from 2022-03-15–2022-04-11, notably:

- `pancreaticoduodenectomy` is a new enumerated value in `RulesOfAcquisition`.
- `higpin` is a new enumerated value in `Precancers`.
- `sarcoma` is a new enumerated value in `Lesion`.
- In class `Biospecimen`, these fields are no longer enumerations or floats, but nullable strings:
- `storage_method`
- `section_thickness`
- `shipping_destination`
- In class `Genomics`, `sequencing_platform` is now a nullable string.

The CDE changelog also mentions that `location_extent_extraprostatic_extension` and `seminal_vesicle_invasion` (although it misspells it "vessicle") have two new enumerated values, however in Sickbay these are simply strings, so there's no need to change anything.

This release contains some incompatible changes to accommodate CDE updates from 2022-01-20 through 2022-01-31 See [the CDE changelog](https://mcl.nci.nih.gov/resources/standards/mcl_cdedictionaries_changelog_v6.xlsx) for details. However, given the moribund nature of the Consortium for Molecular and Cellular Characterization of Screen-Detected Lesions, these changes are simplified for the following reasons:

- The project has concluded and we must intake any form of outstanding data presented (as [Kristen Anton](https://github.com/kristenanton) said: "MCL is ended, we should take what data we can.")
- No one actually writes queries for the relational database or uses the query interface of the ORM.
- The user interface just wants plain text over JSON anyway.

Given that, this and all future changes no longer models 1-to-many relationships as traditional relational database structures of 1-table-row-to-multiple-other-table-rows. Instead, we just use plain text and expect pipe (`|`) separated strings. This is what the input to this database is (pipe-separated cells in a spreadsheet) and what the UI expects, so the Herculean overhead of making the tables and ORM models to support this is ultimately fatuous.

The changes in this release include:

- The enumeration `GroupStage8` now has an additional value, `unknown`. This is not currently reflected in the [Lung CDEs](https://mcl.nci.nih.gov/resources/standards/lung_v2_31jan2022.xlsx) at this time, but we're not concerned.
- In the `LungOrgan` class:
- The following values are now string types instead of enumerations:
- `primary_adenocarcinoma_differentiation_type`
- `prior_treatment`
- The `ajcc_8_lung_pathologic_m` has an error; the enumeration name was duplicated as `metastasis_enum`; it's now `metastasis_enum8`.
- The changelog for CDEs says that all 14 AJCC fields are now optional; but this was already the case for Sickbay since 1.1.0.
- In the `ProstateOrgan` class, the following values are now string types instead of enumerations:
- `location_dominant_nodule`
- `location_secondary_nodule`
- `location_extent_extraprostatic_extension`
- `location_nature_positive_margins`
- `summed_length_positive_margin`
- `seminal_vesicle_invasion`

This release contains some incompatible changes in order to accommodate CDE updates from 2021-08-26 through 2021-11-18. Please [see the CDE changelog](https://mcl.nci.nih.gov/resources/standards/mcl_cdedictionaries_changelog-4.xlsx) for highly pedantic details of these updates. The changes to the software include:

- On class `Organ`:
- `histopathology_precancer_type` was a 1-to-many attribute of `LungOrgan` only; now it belongs to _all_ organs as 1-to-many.
- This base class now has the following optional attributes:
- `ajcc_clinical_m`
- `ajcc_clinical_n`
- `ajcc_clinical_t`
- `ajcc_clinical_stage`
- `ajcc_pathologic_m`
- `ajcc_pathologic_n`
- `ajcc_pathologic_t`
- `ajcc_pathologic_stage`
- `lymph_nodes_tested`
- `lymph_node_location`
- On the class `LungOrgan`:
- There are _numerous_ changes. For one, the `ajcc_staging_system_edition` indicates whether the entire record uses the AJCC Staging edition 7 or 8, and depending on this, it tells which set of attributes to use.
- The attributes are:
- `ajcc_7_lung_clinical_m`
- `ajcc_7_lung_clinical_n`
- `ajcc_7_lung_clinical_t`
- `ajcc_7_lung_disease_stage`
- `ajcc_7_lung_pathologic_m`
- `ajcc_7_lung_pathologic_n`
- `ajcc_7_lung_pathologic_t`
- `ajcc_8_lung_clinical_m`
- `ajcc_8_lung_clinical_n`
- `ajcc_8_lung_clinical_t`
- `ajcc_8_lung_disease_stage`
- `ajcc_8_lung_pathologic_m`
- `ajcc_8_lung_pathologic_n`
- `ajcc_8_lung_pathologic_t`
- Note that all of these attributes are _optional_; this is because it's also possible that `ajcc_staging_system_edition` is `unknown` or `not_reported`, in which case we can't enforce that a specific set of the above attributes are actually used.
- Lungs also have a new attribute: `lymph_nodes_positive`, an optional integer.
- On the class `ProstateOrgan`, these attributes have moved "up" into the superclass `Organ`:
- `lymph_nodes_tested`
- `lymph_node_location`
- `ajcc_clinical_m`
- `ajcc_clinical_n`
- `ajcc_clinical_t`
- `ajcc_clinical_stage`
- `ajcc_pathologic_m`
- `ajcc_pathologic_n`
- `ajcc_pathologic_t`
- `ajcc_pathologic_stage`
- In class `Biospecimen`, these attributes were required and are now optional:
- `days_to_collection`
- `time_excision_to_processing`
- `days_to_storage`
- The following enumerated types have changed:
- `TStage7` no longer includes the terms `t1c` or `t1mi`
- `ClinicalMStage7` has dropped the terms `M1c` and `pM1`
- For `GroupStage7`, the following permissible values are no longer permissible:
- `ia1`
- `ia2`
- `ia3`
- `iva`
- `ivb`
- `Precancers` now includes a `normal` kind
- `Fixatives` now supports a `not_applicable` value
- When it comes to `Storage` you now have two new options
- `room_temperature_then_refrigerated`
- `frozen_at__20c`
- `SlideCharges` has made these values impermissible: `cm0`, `cm1`, `pm1`, `pm1a`, `pm1b`, `pm1c`
- We now finally have a blessed description for `Treatment` instead of the kind contrived by a mere software developer
- At long last an expert has realized that `cannot_be_determine` should be `cannot_be_determined` in `Necrosis`
- The following new enumerations are ready for use:
- `ClinicalMStage8` with 8 values
- `ClinicalNStage8` with 7 values
- `GroupStage8` with 17 values
- `AJCCMetastasisStage8` with 8 values
- Removal of [zc.buildout](http://www.buildout.org/en/latest/). We cannot recommend this tool less. Just use virtual environments like everyone else.

For issue https://github.com/EDRN/MCL-metadata/issues/22

- Additional permissible value on `sequencing_platform` (enum `GenomicAnalyzer`), namely `illumina_hiseq_1500`.
- Changed the `read_length` from numeric to a string (10)
- Note that we do not have schema migrations set up so these steps must be run manually:
- `ALTER TABLE "genomics" ALTER "read_length" SET DATA TYPE CHARACTER VARYING(10)`
- `ALTER TYPE "genomic_analyzier_enum" ADD VALUE 'illumina_hiseq_1500' AFTER 'illumina_genome_analyzer_iix'`