/maxscale/logs/data endpoint returns less rows than expected using maxlog as the source

If maxlog is used as the log source, the /v1/maxscale/logs/data endpoint would return less rows than requested if the log span it covers contained lines with the log footers or headers. This also caused the row numbering to not be sequential which could cause invalid cursor positions to be created.

The fix to this is to use the valid row count as the line numbers, not the actual line numbers in the file.