Initially aggregate loans by appropriate groups.

[jdhoffa]

I was hasty in writing #263, in which I added an aggregation step to aggregate similar loans to different companies, if these loans mapped to the same ALD company. I used a group_by_at(...), and setdiff() to aggregate by all but a few columns, which was lazy of me, and didn't think about the downstream issues of grouping by a set of columns that should in fact map to the same information.

Here, I have made the aggregation step more specific, by grouping only by the relevant columns:

group_by(
      .data$level,
      .data$loan_size_outstanding_currency,
      .data$loan_size_credit_limit_currency,
      .data$name_ald,
      .data$sector_ald
      ) 

Relates to #262
Closes #265

[maurolepore]

LGTM but WARNING: This PR seems to depend on another open PR. Please remember to handle that dependency.

[jdhoffa]

I'm going to close this, and open a new PR request that doesn't depend on #268