Refactor iati_transactions code into a general-purpose reload function that could create multiple tables

At present we have the following data update scripts:

1. iati.sh
2. iati_datastore.sh
3. iati_registry_refresh.sh
4. iati_transactions.sh
5. iati_transactions_retry.sh

We should clean up, simplify, and refactor these scripts such that they:

1. Run Python/iati_refresh.py, marking IATI datasets as either new, modified, or stale in the iati_registry_metadata table.
2. Run a new script based on Python/iati_transactions.py, which is capable of modularly modifying multiple IATI-based tables at the moment new data is loaded (IATI nomenclature would call this iati_reload.py).
3. Run auxiliary scripts such as Python/iati_rhfp.py which rely on the entire data structure and don't need to be run strictly on new/modified files.

The idea is that instead of running iati_transactions.sh run once, at which point the information about whether a dataset is new is destroyed, any number of tables can be progressively built during the reload process just as iati_transactions.py is.

Start by reading iati.sh, iati_datastore.sh, iati_registry_refresh.sh, iati_transactions.sh, iati_transactions_retry.sh, and the referenced Python scripts iati_refresh.py, iati_transactions.py, and iati_rhfp.py. Map their current responsibilities and define the reload flow so registry status remains available while multiple tables are built; done means the scripts are simplified and the new modular reload process supports the listed auxiliary work.

Written by the indexing model from the issue text.