inconsistent use of m

[Martin Albrecht]

Hi,

I want to fix one small inconsistency:

In UniformNoiseLWE the value m is fixed and cannot be overwritten by the user
by providing an m by hand. In LindernPeiker no m is fixed and by default as
many samples as needed can be asked for.

I propose for both classes: if m is None then the default restriction is
implemented. If the user provides a larger m then so be it.

The rationale is that by default it behaves as expected but if one wants to
test how things degrade as more samples are made available this should be
possible.

I will provide such a patch tomorrow morning unless I hear an objection. I
hope that's okay.

Cheers,
Martin

--
name: Martin Albrecht
_pgp: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x8EF0DC99
_otr: 47F43D1A 5D68C36F 468BAEBA 640E8856 D7951CCF
_www: http://martinralbrecht.wordpress.com/
_jab: martinr...@jabber.ccc.de