clarify the meaning of the weights term for TV denoising functions

[grlee77]

Description

In the Zulip chat, there was a question regarding the "weight" term in denoise_tv_bregman.

In that implementation, smaller weights lead to more smoothing whereas the opposite is true for denoise_tv_chambolle.

The underlying cause seems just to be which term in the cost function the weight applies to (i.e. either the data-consistency or the TV regularization term). We should update the docstrings to clarify where the weight is being applied to make its function less opaque.

[gmelon]

Is this issue still valid? If so, I would like to revise it.

[rfezzani]

Any PR is welcome 😉. Thank you @gmelon!