The method in question was doing multiple string concatenations with

+=, which is very inefficient for large number of operations becuase

each operation results in a memory allocation. We now build a single

string with string interpolation for each field and file passed in and

join all the blobs together, which will require only a single memory

allocation for the final string, instead of O(|fields| + |files|)

allocations.

Ran unit tests.