Remove all instances of seek() so that we can handle streaming data again

[pudo]

After some format detection fixes, we now have a few calls to seek() in the CSV module. Those cannot work on urllib-style http request data. One of the main use cases for messytables is to do streaming web data. We should remove these calls, even if this results in a loss of functionality wrt. type detection.

[rufuspollock]

Some thoughts:

• At present urls do not cause a problem because of hack in any.py to make_stream_seekable (which reads entire file into memory). I imagine that hack should go as part of this refactor (?)

Alternative solutions (to removing use of seek):

• implement a urllib style urlopen that supports seek - behind the scenes it could just make multiple urlopen calls, and if full seek support is needed you could use HTTP Range headers (see this example)
• A simple alternative given that all we need is seek(0) is to add some kind of intermediate wrapper around the stream that buffers, say, the first 10k/100k bytes and allows seek within that

That said we only seem to have 2 places seek is used (commas.py) and one place in any.py.

[rufuspollock]

@domoritz is there a status update here - i know you were chatting with @pudo on this (and see you've started some work in a branch).

It would be really nice to get this fixed (in particular it would allow us to upgrade dataproxy which would be super useful!)

/cc @nigelbabu