In the earliest stages of developing the protocol, we did an extra
cleanup before PCR, but we found it added too much variability.
Taking it out simplified the protocol while making the results more
consistent, but still good.
The region table says about 3.2 nM for sample "A+" (though it goes
all the way up to 9000 bp and you might want to cut it off around
700 instead), which is indeed a little low, but it's more than
enough for a sequencing run even if this is the only sample in the
batch. Sample "C+" has some mild overamplification (slow-migrating
fragments) so if anything I would reduce the PCR by one cycle from
there.
There's basically no adapter dimer in sample "C+" but the amount in
"A+" isn't terrible either. If you're still tweaking things and you
were previously using the 0.7X cleanup, you could try switching to
the 0.6X and see what you get, but then the yield might really
become a problem. I would probably just leave it how it is, or use
more input material if you have it.