Update docs to mention trimgalore core usage nuances

[j-andrews7]

When running trimgalore on PE data, alloting 8 or fewer cpus to the task will result in only 1 being used. This dramatically slows down this task in a manner that is not readily apparent. Using 9 or more cpus will run trimgalore in a parallelized manner as expected.

There have been discussions related to why this is necessary here and the trimgalore docs also mention it.

Describe the solution you'd like

Mentioning this in the pipeline docs would be helpful, particularly for users that use custom configs.

[drpatelh]

Added a brief description to the pipeline output docs here :) 6e5ca3e