condition results object changed, new plotting code

[jelmerborst]

... and many associated changes:

• reshaped return Dataset for condition fits. Includes a 'condition' index for magnitudes and parameters, and a 'cond_x_participant' and 'cond' index for eventprobs.
• output of condition fits now has nan's for missing events/stages.
• adjusted compute_times to deal with the new format
• added mean_in_participant argument to compute_times which is by default true: it now first calculates the mean within participant/condition, then across. If false, previous behavior.
• adjusted compute_topologies to deal with new format
• added mean_in_partipant argument to compute_topologies (LRPs seem to come out better)
• both compute_topologies and compute_times return nans for missing stages/events
• changed topo calculation to consider hmp.em_method, included initialized HMP object as argument
• added the option to plot no error bars and standard errors, moved error bar calculation to compute_times
• fixed plot functions wherever necessary
• sliding_event_mags now uses hmp.em_method

[jelmerborst]

It also fixes #95, #76

[jelmerborst]