clustering analysis and other platform paper figure generation code

[matchings]

changes to data_access.loading:

• [x] adds get_stimulus_response_df function that uses mindscope_utilties to get time aligned traces, for cells (dFF, events, filtered_events) or behavior timeseries (running_speed, pupil_width, lick_rate), and appends the annotated stimulus_presentations table (with mean pupil_width, running_speed, trial info, etc)
• [x] updates multi_session_response_df code to use new stimulus_response_df instead of old ResponseAnalysis class
• [x] update various cluster scripts to work with slurm
• [x] adds option to limit_to_closest_active sessions to get_platform_paper_experiment_table to return only last familiar and second novel active (the sessions we are using in platform paper)
• [x] adds option to limit_to_closest_active and limit_to_matched_cells to get_cell_table function, as well as a get_matched_cells_table function that does this automatically
• [x] adds a depth column to experiment_tablethat takes the average imaging_depth across sessions for a given FOV because the imaging_depth can vary slightly across sessions and put a given FOV on opposite sides of a depth threshold if it is close to the threshold (ex: one session is 245 and another is 251 for the same FOV, with a depth threshold of 250)
• [x] adds layer column to experiment_tablethat labels experiments as upper or lower depending on the depth for that FOV (uses average depth across experiments, because imaging_depth varies across sessions for a given FOV)
• [x] sets get_extended_stimulus_presentations option of loading.get_ophys_dataset() to False by default because it takes so long

changes to ophys.response_analysis.cell_metrics:

• [x] fixes error in fano factor cell metric
• [x] adds change modulation index and omission modulation index to cell metrics code
• [x] switch cell metrics code to use data_type and event_type strings instead of use_events and filter_events Booleans to align with new get_stimulus_response_df functions
• [x] adds function to z-score pupil data to data_access.processing.zscore_pupil_data

changes to dimensionality_reduction.clustering:

• [x] adds a whole bunch of functionality to dimensionality_reduction.clustering.processing and plotting
• [x] adds a script to create all clustering plots for figure 4 as dimensionality_reduction.clustering.figures, with the ability to select across session normalized dropout scores, or signed dropouts
• [x] adds code to plot single cell example figures with GLM kernels and dropouts for matched cells to visualization.ophys.platform_single_cell_examples

and finally, the original purpose of this PR:

• [x] adds all Marinas other platform paper figure plotting functions to visualization.ophys.platform_paper_figures

[alexpiet]

Both CircleCI tests are failing due to a Markdown package change: https://github.com/aws/aws-sam-cli/issues/3661

Seems like an easy enough fix

[yavorska-iryna]

@alexpiet Thank you for figuring it out. As far as I understand Markdown package is a part of PyPi. When you say, easy to fix, what kind of fix are you suggesting? Thanks.

[alexpiet]

@yavorska-iryna the github issue I linked had a work-around that is just changing the version of some package used by the circle ci test