Fix read_wi

[Rafnuss]

While working on read_wi, I noted a few minor update of the WI standard:

• images.csv -> images_xxxxxx.csv file name change
• new camera_name (free text) field in cameras and deployements
• new remarks (free text) field in deployments
• new behavior (free text) and bounding_boxes (see below) in images.
• Issue with different type of human

Here is an example of bounding_boxes:

{"{\"detectionBox\":[0,0,0.873758137,0.736430466]}","{\"detectionBox\":[0.274836838,0.402429402,0.816743791,0.935070515]}","{\"detectionBox\":[0.183905229,0.207098484,0.834438,0.924319863]}"}

As I understand it, the taxonomy is actually based on the scientificName (rather than taxonID??). It lumps my different type of human together. WI offers these type of humans:

• Horseback Rider 2d1bb69a-285a-4934-8cdb-a0355995d98b
• Human-Faces e2f84cfb-5976-4adf-a91d-4953c2916b31
• Human-Camera Trapper 066c4187-4a03-4531-9f62-ac585ad0a138
• Human-Hunter f5ec47a5-e7a4-4158-a0b4-a6481ae3f06f
• Human-Researcher e6f88ed1-c305-4e94-b53d-8eec1b1d980a
• Human-Tourist 795880f2-2677-41bf-be13-72c58056f08d
• Human-Park Ranger 0ffb7a18-d6df-454c-a014-c99f52ec22a1
• Human-Resident 128d0fac-f96d-4a98-af18-2903dbbe1568
• Human 990ae9dd-7a59-4344-afcb-1b7b21368000
• Human-Maintenance crew 41ebc3f1-ecc7-497c-acd4-6b35623a9465
• Human-Pedestrian 0de272bf-5eab-4198-a888-3a6e92f1460d
• Human-Soldier 36fc2d39-cda9-47a1-b285-98334231a034
• Human-Biker c491e45f-4440-4bac-8013-7037262df4ea

Is there a way to keep them?
For now, I've fixed this with this code, but it isn't great!

  # Add human type in scientific name to avoid lumping them
  wi_images <- wi_images %>% 
    mutate(
      species = ifelse(
        species == "sapiens",
        glue::glue('{species}-{sub("Human-", "", common_name)}'),
        species)
  )