- change a*b to a$\times$b (in LaTeX mode)
- don't show *'s
- shorten Piracy(SR)/Policy(SR) to Pir.(SR)/Pol.(SR)

- in the general latex tabulation, use a minipage instead of a
  \nopagebreak; the latter only prevents page breaks at once location,
  but does nothing about page-breaks in the title that appears right
  before it (so half the equation could end up at the bottom of a
  page/column, and the other half at the top of the next page/column).
  Putting the whole thing in a minipage fixes that.

- Don't put a \n before the equation: under latex, it becomes a \\, but
  the \\ is in an invalid location and generates "no line to end here"
  errors.

When making the graph small (e.g. to be able to fit it in the paper)
the end-point tick values overlap the regular tick values, so just
suppress them instead (the non-round number was a bit ugly anyway).

The code failed (because the matrix didn't have enough columns), and
short-run variables weren't properly displayed anyway.

It says "the default is 6" when it was actually 5 (but it also isn't
needed since the default is shown before the description anyway).

New flags --initial and --pre can be used to include them, if desired.

It isn't needed after all; the bug was actually a debian ycmd problem
(debian bug 825905).

(until https://github.com/Valloric/YouCompleteMe/issues/2330 is fixed).

When we have all the values we need, there's no point in holding a lock
for the first simulation to load because there's nothing to infer; so
set the periods/piracy_begins/policy_begins in the main thread and
propertly set need_parameters = true only if one (or more) is omitted;
otherwise inference is not needed.

It's an incredibly boring series: 0 until the policy begins, then a
constant (of course there's variation in the constant across
simulations, but not over time).

Also renamed books_written to books_written_pc to reflect that it's a
per-capita (actually per hundred readers) value.

Added (commented-out) stubs for reader_fines and readers_caught_pc;
these need deeper changes (and a simulation re-run) as they aren't
currently tracked in the creativity state file.

Exiting in preload mode wouldn't trigger the thread count decrement;
fixed it by breaking the while loop instead of returning directly.

Mostly copied from creativity-data.

prctl needs to run in the main thread to show up in top (since top
shows the main thread's name when not showing individual threads),
this commit adds machinery to handle that via a wait loop in the main
thread.

Also a minor fix to make --help-series work without -o

The progress is displayed both in the error output and in the process
name shown in top (via prctl).

books_written, books_bought, books_pirated, and books_public are all not
particular useful as total numbers because, ceteris paribus, a larger
simulation will have larger values of all of them.

This changes them all to per-capita values (the first changes to a
per-100-readers value, the rest to per-reader averages).

These can extract the list of simulations that will be used for
analysis (so that these could then be used to generate series files).

- policy_public_voting_tax, _votes were missing
- policy_catch_cost was missing
- policy_catch_fine[3] wasn't valid

There are still problems here: this fixes enough for the data summary
to support policy filtering.

Manually playing with bits in various places is dirty: this adds a new
Policy class that abstracts all the bit values, while providing various
nicer operations and methods related to a policy value.

This also includes a eris::serialize::serializer<Policy> implementation
(which is just a simple wrapper around a serializer<uint32_t>) so that
a Policy object can be directly serialized.

The former is cumbersome.