Currently, when running tests with coverage, where the data produced by coverage report is large, each reporter is most probably running a row-by-row query from client to DB.

We should prevent this as it significantly impacts the overall execution time of run when done va utPLSQL-cli.

Ideally, data from reporters that do not report to screen but do reporting in background , should be read in bulks of 100, 1000 or even 10000 rows.

If you have a look at our test runs on Travis,
~50 seconds is time for utPSLQL unit test run, while overall end-to-end run takes ~80 seconds.
I can imagine that on really large schema, most of the time might be spent on reading coverage report by cli.