Changed light_side policy and added albedo functions

Now get_light_side returns the actual light side of the closure regardless of the singularity. And I have added albedo functions to all the bsdf's to make sampling easier and better. Note that those albedo functions don't have to be accurate.

Although they are in the same review, both things will be checked in as separated commits.

[larrygritz, 2010-03-25 22:54:55]

http://codereview.appspot.com/756041/diff/1/2
File src/include/oslclosure.h (right):

http://codereview.appspot.com/756041/diff/1/2#newcode258
src/include/oslclosure.h:258: /// be used for sampling decissions
decisions.

Can you elaborate more?  I think somebody reading this comment will have no idea
what this is for, or why it's ok to only be an approximation.  You say it should
be under 1.0, but some of the implementations return 1.0.  Also, you should
explain here that things that we don't know how to sample ought to return 0.0
(I'm surmising, looking at some of the subclass implementations).

[aconty, 2010-03-25 23:30:48]

Addressed comments

[larrygritz, 2010-03-25 23:36:22]

thanks, that's much more clear now

[ckulla, 2010-03-26 16:57:12]

LGTM2

http://codereview.appspot.com/756041/diff/15001/16001
File src/include/oslclosure.h (right):

http://codereview.appspot.com/756041/diff/15001/16001#newcode257
src/include/oslclosure.h:257: /// be less or equal than 1.0 and it is not
guaranteed to be accurate.
"less than or equal to 1.0"

[aconty, 2010-03-26 17:43:38]

> guaranteed to be accurate.
> "less than or equal to 1.0"

Thanks, fixed that. Will commit when the render part gets approved