Update command-line option defaults for khmer 2.0

[ctb]

• abundance-dist-single: -t and --csv should become default
• abundance-dist: --csv should become default
• count-median.py: --csv should become default
• count-overlap.py: --csv => default
• filter-abund-single.py: -t => default
• load-into-counting.py: -t => default, maybe summary info too?
• split-paired-reads, -p => default (see split-paired-reads.py has nonstandard behavior with orphaned reads #847)

[ctb]

(It's not just about updating defaults, actually -- there should be no option to turn off --csv or -t, in fact.)

[ctb]

• look into readstats

[ctb]

• normalize-by-median report_fp change.

[ctb]

Partial dup of #1011, which is now closed.

[ctb]

What wasn't fixed in #1180 is fixed in #1164.