Clarify, implement and document the workflow for manual deduplication

[LukasWallrich]

Currently, CiteSource allows to identify potential duplicates, but not to deal with them - need to figure out how to reexport the relevant ASySD functions.

@TNRiley can you check if ASySD::dedup_citations_add_manual works with our dedup_citations() output?

[TNRiley]

walking this through step by step. Including numbers based on the data in the new_benchmark_testing vignette so it's easy to walk through and for me to remember

If we run CiteSource::dedup_citations(raw_citations, manual = TRUE) we return a list

• unique_citations_manual$unique (499 rows - which represents the 499 distinct records across all sources)
• unique_citations_manual$manual_dedup (2 rows - which represent 2 different sets of potential duplicates)

Then if we run CiteSource::dedup_citations_add_manual (unique_citations_manual$unique, unique_citations_manual$manual_dedup)
(Currently, there is an issue with the extra_merge_fields argument, but if we remove the extra_merge_fields = "cite_string" argument from the CiteSource::dedup_citations_add_manual function) we return

• a single dataframe with 497 rows (this is due to the fact that ASySD and CiteSource equate the additional_pairs as TRUE duplicates)

Therefore, a user to deal with potential duplicates within CiteSource (in R at least - the shiny is okay), we would need to build in a function or develop a table to show the user the potential duplicate data, record their decision, and assign any TRUE duplicates that were identified as the data that is then passed through dedup_citations_add_manual's additional_pairs.