[BUG] Vr binning not accurate around track limits

[HDClark94]

Describe the bug
Plotting the speeds_binned_in_time from processed_position_data shows speed profiiles that no not match with that shown in the behaviour subfolder. This was highlighted by Nina Kudryashova

processed_position_data pos_binned_in_time vs speeds_binned_in_time

speed histogram from /MountainSort/Figures/Behaviour

Running environment

• Data in datastore: e.g. Harry/test_recording/M3_D28_2020-12-05_15-05-24
• Branch: master
• Platform: Eleanor

To Reproduce

    fig, ax = plt.subplots(1,1, figsize=(6, 6))
    for i in range(len(processed_position_data)):
        pos_binned_in_time = processed_position_data["pos_binned_in_time"].iloc[i]
        speeds_binned_in_time = processed_position_data["speeds_binned_in_time"].iloc[i]
        ax.scatter(pos_binned_in_time, speeds_binned_in_time, c="blue", alpha=0.1)
    plt.ylabel('speed cm/s', fontsize=30, labelpad = 10)
    plt.xlabel("Location (cm)", fontsize=30, labelpad = 10)
    plot_path = output_path + '/speeds_from_processed_position_time_binned.png'
    plt.savefig(plot_path, dpi=300)

I am fixing this in the branch improve_vr_binning

[HDClark94]

Shown here is the same plot using branch improve_vr_binning