implement the perform_matching_calculation

it's currently implemented as a dummy function in the worker process.

it receives

* chain_id
* round_id
* algorithm

steps:

* fetch round, applications, and donations data from the indexer
* maybe stream it to a temporary file to avoid having too much data in memory but TBD during the implementation
* call the specific algorithm function passing data and returning the final results file url

Start in the worker process at the dummy perform_matching_calculation function and trace how chain_id, round_id, and algorithm are passed in. Review the indexer access for round, applications, and donations data, then determine how the selected algorithm receives that data. Done means the calculation fetches the required data and returns the final results file URL.

Written by the indexing model from the issue text.