Documenting changes from DF v1 to DF v2 (refactored acquisition)

Starting a place to document things as they come up.

To close:
- [ ] Finish documenting all changes
- [ ] Move to a permanent location here: https://github.com/AllenNeuralDynamics/aind-behavior-blog/tree/main/docs/dynamic-foraging

# How did we verify the new data
CO capsule where we compared old and new data https://codeocean.allenneuraldynamics.org/capsule/2699857/tree

# Trials table

### Columns removed for acquisition v2

|Status | Column |Notes|
|- |-|-|
|Removed | auto_train_curriculum_name | AutoTrain no longer exists. Curriculum name in metadata |
|Removed | auto_train_curriculum_schema_version | AutoTrain no longer exists. check in metadata|
|Removed | auto_train_curriculum_version |AutoTrain no longer exists. check in metadata |
|Removed | auto_train_engaged | AutoTrain no longer exists. Need to clarify, but I believe we can't override|
|Removed | auto_train_stage |AutoTrain no longer exists. check in metadata |
|Removed | auto_train_stage_overridden | AutoTrain no longer exists. Need to clarify, but I believe we can't override|
|Removed | reward_outcome_time | Internal variable, this is NOT choice time or reward time. No reason to save|
|Removed | reward_random_number_left | Outcome of random number draw, no reason to save |
|Removed | reward_random_number_right | Outcome of random number draw, no reason to save |
|Removed | side_bias_confidence_interval | Acquisition software does not compute confidence interval. It can be computed with this function [compute_side_bias()](https://github.com/AllenNeuralDynamics/aind-dynamic-foraging-basic-analysis/blob/c83783ff74e0757a52380e31952678972a5d7a3b/src/aind_dynamic_foraging_basic_analysis/metrics/trial_metrics.py#L89)|

### Columns modified for acquisition v2

|Status | Column |Notes|
|- |-|-|
| modified| animal_response | Previously was a float (0.0, 1.0, 2.0), now it is an int (0, 1, 2)|
| modified| base_reward_probability_sum | This is NaN when the blocks are uncoupled, previously the value was set but ignored|
| modified| block_beta| This is NaN when the blocks are uncoupled, previously the value was set but ignored |
| modified| start_time| This used to refer to the start of the ITI period, in acquisition v2 sessions it refers to the `quiescent_start_time` |
| modified |stop_time| This used to refer to the end of the reward consumption period, in acquisition v2 sessions, it refers to `ITI_stop_time`|
| modified | delay_beta | This is NaN because the delay period is fixed to be of length 1|
| modified | delay_max | This is NaN because the delay period is fixed to be of length 1|
| modified | delay_min | This is NaN because the delay period is fixed to be of length 1|
| modified | bait_left | In acquisition v1, this column uses "bait" to mean "if the mouse licks, it will get a reward" that reward could be given by a reward being generated on that trial (the normal "earned" reward sense), or from baiting on a previous trial. In acquisition v2 this column only refers to the baiting from a previous trial, as reward outcomes for a single trial are not saved. |
| modified | bait_right | See description in bait_left |

### Columns added for acquisition v2

|Status | Column |Notes|
|- |-|-|
| added| ITI_start_time | Time when the ITI interval starts |
| added| ITI_stop_time | Time when the ITI interval stops - this is now redundant with `stop_time`|
| added| quiescent_start_time | Time when the quiescent interval starts - this is now redundant with the backwards compatible `delay_start_time` and `start_time`|
| added| quiescent_stop_time | Time when the quiescent interval stops |
| added| response_start_time | Time when the response interval starts |
| added| response_stop_time | Time when the response interval stops |
| added| reward_consumption_start_time | Time when the reward consumption interval starts |
| added| reward_consumption_stop_time | Time when the reward consumption interval stops |
| added| anti_bias_left_water | Was there anti-bias autowater on this trial? |
| added| anti_bias_right_water | Was there anti-bias autowater on this trial? |
| added| anti_bias_lickspout_movement | Was there antibias movement on this trial? |

> [!IMPORTANT]
> The columns `rewarded_historyL` and `rewarded_historyR` only annotate earned rewards. Auto-water and manual rewards are not listed in rewarded_historyL/R. Users should inspect the columns `auto_waterL`, `auto_waterR`, `anti_bias_left_water`, and 'anti_bias_right_water`. These columns are summed in `extra_reward` which is added by [data-utils.create_df_trials](https://github.com/AllenNeuralDynamics/aind-dynamic-foraging-data-utils/blob/8002fce1dfd4298f37fd6af4a233fcfcc77a40b6/src/aind_dynamic_foraging_data_utils/nwb_utils.py#L313).

> [!IMPORTANT]
> For timing events (response, reward_consumption, and ITI), the columns with the suffix `_duration` define the parameter value for how long that period could be. The columns with suffix `_stop_time` and `_start_time` document the actual time in which these epochs started and stopped. The actual duration between `_start_time` and `_stop_time` will not necessarily match the `_duration`. For example, the response period will be shorter than `_duration` if the mouse makes a response. The reward_consumption period will differ by clock jitter and software delays.