Replace `_numpy_allclose` in `cf.aggregate` and remove

During discussion today we noted that _numpy_allclose will (after #254) no longer be required by the Data modules, being replaced by efficient Dask logic in a new function in dask_utils.

Since the only remaining usage of it is in the aggregate module, we should replace that one use with potentially consolidated or more specific logic based on the requirements in that context, then remove _numpy_allclose. Potentially Data.equals or similar could be used to achieve this, but it remains to be seen.

(Raising as an issue because we are focusing on the Data side, so this isn't priority.)

Iniziate da _numpy_allclose in cf/functions.py e tracciatene l'uso rimanente nel modulo aggregate. Leggete i requisiti di aggregazione circostanti e la sostituzione di dask_utils a cui si fa riferimento, quindi confermate che la logica specifica del contesto funzioni e che _numpy_allclose non abbia più utilizzatori.