Fixed some errors and tried to improve usability

[marms7]

Summary

I was playing around with this code and noticed some things that seemed miscopied from whatever code was used to generate the figures in the associated journal article. I figured I would share my changes in case it's useful to the next person.

Changes

• The first time coeff was defined, one of the inputs to the polyfit function was TE_modes indexed in one dimension by second_level_comp, which was blank. Indexing by a blank array was giving an error, so I changed the input to be TE_modes.
• When the component identify function was run, it was being assigned to an array of the wrong dimension (as a variable for reject_comp_1 was excluded). I added this variable.
• nilearn._utils is an outdated library so I updated it to nilearn.image.
• Fixed typo "ploting" changed to "plotting".
• Number of echoes was hardcoded which made it difficult to use for new data. Created an echo_times variable and added it as an input to the component_identify function. This variable is used within the function to set additional variables num_echoes (number of echo times in the echo_times variable) and interp_degree (number of echoes - 1). Note that the degree of the polynomial was hardcoded but possibly incorrectly set to 2 (rather than 3) in the original code.
  • There are some hardcoded thresholds, it is unclear if any should change with number of echoes. I kept them as is.