Prevent massive read of data

[alexandre-lecoq]

Reproduction steps

• Generate an excel file using a query which will read 2000000 rows over several databases.

Observed

• It will take hours to read the data, and hours to merge the data. The result might be 9 GB of data. Garbage collection will take hours. It won't be possible to get an excel file in a reasonable time, if at all.

Expected

• Prevent doing things that can't work.

How to fix

• Hard limit the size of tables at every step (reading, merging, generating). If a table is bigger than the hard limit : drop the table and issue an error log.
• Add 2 parameters --maximumTableRows and --maximumTableColumns to reduce the limit to a certain amount.
• The limit cannot be increased.
• The limit should not apply when results are discarded using --discardresults

The hard limit should be 1048576 rows and 16384 columns per excel specifications.
See https://support.office.com/en-us/article/excel-specifications-and-limits-1672b34d-7043-467e-8e27-269d656771c3

If #48 is implemented those fix would be questionable.

[alexandre-lecoq]

Since CSV exporter #48 was implemeted. It should be thought again. CSV exporter is kind of ment to export data volumes that ExcelExporter cannot export.