557 fix first instance issue for Lime timeseries

[geek-yang]

Fix #557.

Now it ensures the first instance is the original data, which satisfies the requirement from https://github.com/marcotcr/lime/blob/fd7eb2e6f760619c29fca0187c07b82157601b32/lime/lime_base.py#L148-L149

[cpranav93]

It looks great! One point of modularity, it might be better to have the "masks[0,:,:] = 1.0" as a flagged option in maskers.py rather than in lime. But we can also do that later once Chris is finished with the masks.

[geek-yang]

It looks great! One point of modularity, it might be better to have the "masks[0,:,:] = 1.0" as a flagged option in maskers.py rather than in lime. But we can also do that later once Chris is finished with the masks.

Good suggestion! I mark it in issue #556. When Chris finished the segmentation, we can implement it. Thanks for your review @cpranav93 😄.