Add T015–T017 to clean three ADB ATO data sets

[HannaMorde]

Data cleaning of "ATO Workbook (TRANSPORT ACTIVITY & SERVICES (TAS))2023.xlsx" was done and merged the output into the cleaned "master dataset.csv".

[khaeru]

Hi @hannamm —thank you very much for your contributions to the iTEM Open Data code! We appreciate that members of the community, such as yourself, make efforts to expand the scope and quality of the data included in this shared resource.

In this review, I will do a few things:

1. Summarize what I think is being added by the commits on this branch. (This is information that should go in the description of the pull request. If I am guessing incorrectly, please correct me by providing the correct explanation.)
2. Identify TODOs to merge this branch/PR.
3. Identify TODOs for further clean-up (after the branch is merged).

To be clear, I understand from you and @soniayeh that the course in which you did this work is coming to an end, so you may not have dedicated time to do (2) or (3). So those TODOs are mentioned without saying who will do them or when—it could be you, me, or someone else; sooner or later.

What is added

• Three data-cleaning modules: item.historical.{T015,T016,T017}.
• Each cleans one data flow from the Asian Development Bank (ADB) Asian Transport Outlook (ATO); I guess the 2023 edition:
  • T015 → TAS-FRA-005(3)
  • T016 → TAS-VEP-005(1)
  • T017 → TAS-VEP-020

TODOs to merge the branch

• Deduplicate the code in the 3 added modules. For instance, the files T015.py and T016.py are about around 420 lines, but >300 of these lines are identical in both files; this makes it difficult to identify which steps are common versus specific to each input data flow. One way to achieve this would be to define the class AtoWorkbook in a single file, then import it in each of the 3 modules, setting some different parameters or creating a subclass.
• Provide a single entry-point in each module. Existing data modules like T000 do not execute any code when they are imported; they instead provide a function like process() that triggers processing of the data.
• Create a pull request in transportenergy/metadata. I see this branch has updated the contents of the Git submodule at item/data, and then updated the pointer to a new commit. This is the correct approach—very good! However, in order for those added contents (I guess including files like "T015_input.xlsx") to be available, they must be pushed to a branch visible in GitHub; and then to be permanent they must be PR's into the main branch of transportenergy/metadata.
• Add documentation, for instance in the docstring at the top of each module. Explain, at least, which specific data are used: at what URL can the original input data be found? This will also clarify which edition of the ATO is handled by the code.
• Ensure the tests pass. This can be started by rebasing the branch on transportenergy:main.

TODOs for follow-up

• Consider overlap with the transport_data.adb code maintained by the Transport Data Commons, here; make use of the TDC code where possible, and keep the features specific to iTEM Open Data.
• Consider overlap with e.g. the class ItemTransformer added in Add T018–T020 to clean three ADB ATO data sets #80; merge common actions to common code.
• Add the new modules to the test suite. item.tests.test_historical contains tests of the existing (T000–T010, T012) data modules, and these tests should be expanded to test the modules added in this PR.