"    Input:  {
               data_uoa                      - experiment data UOA (can have wildcards)
               (repo_uoa)                    - experiment repo UOA (can have wildcards)
               (experiment_repo_uoa)         - use it, if repository is remote
               (remote_repo_uoa)             - if repo above is remote, use this repo on remote machine

               (module_uoa)
               (tags)                        - search by tags

               (point)                       - point (or skip, if there is only one), can be of format UID-
               (subpoint)                    - subpoint (or skip, if there is only one)

               (repetitions)                 - statistical repetitions (default=4)
               (pipeline_update)             - customize pipeline with this dict (useful to update already prepared pipeline from file)

               (dims_to_check) or (dims)     - list of dimensions to check, can have wildcards (if empty, check all); 
                                               alternatively can use a string (useful for CMD)
               (end_of_dims_to_check)        - list of endings of dimensions to compare ...

               (threshold_to_compare)        - % threshold to decide whether dim is different or not

               (record_original_flat_json)   - file to record flat dict with original results
               (record_reproduced_flat_json) - file to record flag dict with reproduced results

               (skip)                        - if 'yes', do not perform comparison

               ==============================
                 If one would like to compare results with another point (for example, to check speedups 
                 rather than just reproducing a given point)

               (compare_data_uoa)
               (compare_repo_uoa)
               (compare_point)
               (compare_subpoint)

               ==============================
                 Prunning (reducing complexity)

               (prune)                          - if 'yes', replay and prune choices!
               (reduce)                         - the same as above
               (reduce_bug)                     - reduce choices to localize bug (pipeline fail)

               (prune_md5)                      - if 'yes' and compilation is used in a pipeline (workflow), check if binary MD5 doesn't change
               (prune_invert)                   - if 'yes', prune all (switch off even unused - useful for collaborative machine learning)
               (prune_print_keys)               - list of keys from flat dict to print during pruning (to monitor characteristics, for example)
               (prune_invert_do_not_remove_key) - if 'yes', keep both on and off keys (to know exact solution)

               (prune_conditions)    - conditions on results (see "ck check math.conditions --help")
               (condition_objective) - which objective to use for characteristics (#min, #max, #exp, ...)

               (solutions) - prune first solution

               ==============================
               (record_uoa)                       - (data UOA or CID where module_uoa ignored!) explicitly record to this entry
               (record_repo)                      - (repo UOA) explicitly select this repo to record

               ==============================
                 Some productivity keys specifically for autotuning pipeline (ck-autotuning repo):

               (local_platform) or (local)   - if 'yes', use parameters of a local platform (to retarget experiment)
               (skip_target)                 - do not select target machines (to customize via host_os/target_os/device_id)
               (no_deps)                     - if 'yes', do not reuse deps
               (skip_clean_after)            - if 'yes', do not clean program pipeline after execution 
                                               (keeping low level scripts in tmp directory for low-level debugging)
               (all)                         - print all comparisons of keys (not only when there is a difference)

               (skip_scenario_keys)          - skip scenario detection (otherwise get keys from the detected module) 



            }

    Output: {
              return       - return code =  0, if successful
                                         >  0, if error
              (error)      - error text if return > 0

              output of a given pipeline

              different_dims - list of different dimensions
            }
"