compaction: base compaction throughput on amount of data read

Today, we base compaction throughput on the amount of data written,
but it should be based on the amount of input data compacted
instead, to show the amount of data compaction had to process
during its execution.

A good example is a compaction which expire 99% of data, and
today throughput would be calculated on the 1% written, which
will mislead the reader to think that compaction was terribly
slow.

Fixes scylladb#14533.

Signed-off-by: Raphael S. Carvalho <raphaelsc@scylladb.com>

compaction/compaction.cc

@@ -788,7 +788,7 @@ class compaction {
         log_info("{} {} sstables to {}. {} to {} (~{}% of original) in {}ms = {}. ~{} total partitions merged to {}.",
                 report_finish_desc(),
                 _input_sstable_generations.size(), new_sstables_msg, utils::pretty_printed_data_size(_start_size), utils::pretty_printed_data_size(_end_size), int(ratio * 100),
-                std::chrono::duration_cast<std::chrono::milliseconds>(duration).count(), utils::pretty_printed_throughput(_end_size, duration),
+                std::chrono::duration_cast<std::chrono::milliseconds>(duration).count(), utils::pretty_printed_throughput(_start_size, duration),
                 _cdata.total_partitions, _cdata.total_keys_written);
 
         return ret;