CSV JMH benchmark file contains an unexpected double value

On one of the PERF axis CI runs a JMH benchmark csv file contained a double value in the "number of samples" column. The code in io.narayana.perf.jmh.BMParser expects an integer so fails with NumberFormatException (see linked CI job)

I don't know why JMH would use a double here (since in all other perf runs for the same benchmark it has used an integer). Anyway, I think I will just update the parser to accept a double or an int for this column.