Fixes various corner cases and other bugs regarding lines with
comments, having data but no endpoints, etc.

Things are now block-focused, rather than line-focused.  This should
give a pretty big speedup to inserting client data, especially when
inserting preformatted data.

Still needs work.

Conflicts:
	nilmdb/server/bulkdata.py
	nilmdb/server/server.py
	nilmdb/utils/__init__.py

This should more or less complete the rocket interface.

Before, a tuple was crammed into args[0].  Now, the three arguments are
args[0:2].

The server buffers the string and passes it to nilmdb.  Nilmdb passes
the string to bulkdata.  Bulkdata uses the rocket interface to parse
it in chunks, as necessary.  Everything gets passed back up and
everyone is happy.

Currently, only pyrocket implements append_string.

Use a new helper, nilmdb.utils.time.float_to_time_string().
This will help if we ever want to change representation (like using
uint64 microseconds since epoch, which saves us from having to
waste bits on the floating-point exponent)

Requires fallocate(2) support with FALLOC_FL_PUNCH_HOLE, as
well as a filesystem that supports it (in Linux 3.7,
tmpfs, btrfs, xfs, or ext4)

There isn't much left in File, so let's move as much as possible
over to C.

Replaces custom tools.allow_methods which didn't return the Allow: header.

This is mostly a matter of taste, but it matches more closely with the
old way that prep did it, and it's more consistent.  It should roughly
match the available precision of floats and doubles.

This is maybe 2.5-3 times faster than the list-based code, which
still isn't amazing, but is decent.

Currently, only append_list is written (and hardly tested)

This is so tests can be run without interfering with a normal server.