import: add csv debug option

[arnetheduck]

This new option saves a CSV to disk while performing import such that the performance of one import can be compared with the other.

This early version is likely to change in the future, hence it's also hidden behind a debug flag.

Here's a sample output:

block_number,blocks,txs,gas,time
8193,8192,0,0,964623666
16385,8192,0,0,1052502040
24577,8192,0,0,1033609541
32769,8192,0,0,1053529132
40961,8192,0,0,1129258484
49153,8192,1514,0,1497880436
57345,8192,3916,0,2453669141

This can be used to generate a bps/tps graph like so (, where the initial ~100k blocks have been cut out since they're mostly empty:

This is generated with gnuplot:

set datafile separator ","
plot 'xx.csv' using 1:(1/(($5/1000000000)/$2)) with lines title 'bps', '' using 1:(1/(($5/1000000000)/$3)) with lines title 'tps'