FInd a way to load less data from files

Loading data from files takes a lot of time (finding location, getting path right).

It would be good to reduce that.

One solution would be to reuse the same data in all exercises, so that it is loaded only once.

Another would be to provide some of it directly in template exercise files.

Start by reviewing the exercise templates and the existing file-loading paths to identify where data is located and loaded repeatedly. Compare the two approaches proposed in the issue and confirm the intended scope with a maintainer. Done means the exercises require less repeated file-location work and load shared data no more than necessary.

Escrita pelo modelo de indexação a partir do texto da issue.