Remove dependencies on dplyr and tidyr

[BrianLang]

This PR removes dependencies on dplyr and tidy from metalite.ae The hard dependencies were removed from the following functions:

• avg_event
• avg_duration
• fmt_pval

Additionally, there was need to update avg_event and avg_duration.

• These functions have been updated to return their output with the same row order as what was input. Row order should depend on order of AEDECOD / par parameter, and this order needs to be preserved in the summarization process.
  • This affected the snapshots of ae_specific, since that is the function with uses ave_event and avg_duration.
• I've also adapted testing and added some extra tests to test that ordering is maintained.

Also I removed the need for dplyr in the function examples.

[BrianLang]

Well, now we know it screwed other things up. Will look.

[BrianLang]

So upon digging in, I found that these functions avg_event and avg_duration do not do any controlled sorting of the output. While you would have expected the output to be in the same sort order as the input, or perhaps an alphabetical sort order, this was not the case. I've updated these functions and their testing to maintain sort order of output to be consistent with the input par argument.

The function extend_ae_specific uses these functions but has not been tested in a way that would uncover this issue.

[BrianLang]

This looks good to me. Thanks for reducing the number of hard dependencies!

Would be happy for some feedback on my approach and implementation if you can find time.

[nanxstats]

This looks good to me. Thanks for reducing the number of hard dependencies!

Would be happy for some feedback on my approach and implementation if you can find time.

Sure, I've scheduled some time next week.