User provided proposal distribution for insert and remove moves

Possibility of providing a user-defined proposal distribution for the insert and remove moves, with the new parameters hist_insert, hist_remove and nbins_histo. The proposal distribution is then skewed with respect to $\Delta \tau=\tau - \tau^{\dagger}$ for the insert move and $\Delta \tau=\tau^{\dagger}-\tau$ for the remove move (following the same conventions as the histograms of the performance_analysis feature).

The user provides two dictionaries hist_insert and hist_remove, with a key for each block name. For each key, the value is a list of size nbins_histo, corresponding to a proposal probability for the insertion/remove of two operators for each bin of $\Delta \tau$. (For the definition of the bins, I follow the same conventions as the histograms of the performance_analysis feature, so the first and last bins are half-sized).

An usage example to build these proposal distributions, using the knowledge of a previous run, would be:

nbins = 100
S.solve(...,performance_analysis=True, nbins_histo=nbins)
hist = S.performance_analysis
hist_insert = {}
hist_remove = {} 
for block_name in list_of_block_names:
    hist_insert[block_name] = hist["insert_length_accepted_" + block_name].data / hist["insert_length_proposed_" + block_name].data   
    hist_remove[block_name] = hist["remove_length_accepted_" + block_name].data / hist["remove_length_proposed_" + block_name].data  
S.solve(...,hist_insert=hist_insert, hist_remove=hist_remove, nbins_histo=nbins)

TRIQS / cthyb

User provided proposal distribution for insert and remove moves #176