Pycmtensor

Fix

- **scheduler**: fix missing args in input parameters
- **scheduler**: fix constantLR missing input paramerer

Feat

- **python**: update to python 3.10

Fix

- **tests**: update tests files to reflect changes in biogeme removal

Feat

- **deps**: remove Biogeme dependencies

Fix

- **expressions**: remove Biogeme dependencies
- **database**: remove dependencies of Biogeme
- **debug**: remove debug handler after each run to prevent duplication
- **models**: add function to return layer output -> get_layer_outputs()
- **debug**: disables tqdm if debug mode is on and activates debug_log

Refactor

- move elasticites from models to statistics for consistency

Feat

- **models**: add functionality to compute elasticities of choice vs attribute in models.py

Fix

- **results**: remove unnessary `show_weights` option in Results
- set default max_epoch on training run to adaptive rule
- print valid config options when invalid options are given as args to train()
- **scheduler**: modified cyclic_lr config loading sequence to fix unboundError
- **train**: turn saving model off for now
- **config**: generate os dependent ld_flags

Refactor

- **utils**: refactored save_to_pickle and disables it

Perf

- **IterationTracker**: use numpy array to store iteration data

Feat

- **models**: Implement the ResLogit layer

Fix

- **config**: set default learning schedule to ConstantLR
- **config**: set default seed to a random number on init