Improve printing

[BjarkeHautop]

The old printing printed a lot, but didn't really show much. Redesigned the printing of Knowledge and Disco objects, what do you think @ekstroem? I can then use this in the JSS paper instead of having to use plots every time.

While redesigning I noticed you couldn't use + syntax in knowledge() for infix operators, such as
A + B %-->% C + D, and I fixed this too. To test this PR locally run

pak::pak("https://github.com/BjarkeHautop/causalDisco")

Showcase of output:

suppressPackageStartupMessages(library(causalDisco))
data("tpc_example")

# Only edge knowledge
kn <- knowledge(
  tpc_example,
  c(child_x1, oldage_x6) %-->% c(child_x2, youth_x3),
  child_x2 %!-->% c(youth_x3, oldage_x5)
)
kn
#> <Knowledge: 6 vars | 4 required | 2 forbidden>
#>   vars: child_x1, child_x2, oldage_x5, oldage_x6, youth_x3, youth_x4
#>   child_x1 %-->% child_x2 + youth_x3
#>   oldage_x6 %-->% child_x2 + youth_x3
#>   child_x2 %!-->% oldage_x5 + youth_x3

# Only tier knowledge:
kn2 <- knowledge(
  tpc_example,
  tier(
    1 ~ starts_with("child"),
    2 ~ starts_with("youth"),
    3 ~ starts_with("oldage")
  )
)
kn2
#> <Knowledge: 3 tiers | 6 vars>
#>   tier(1): child_x1, child_x2
#>   tier(2): youth_x3, youth_x4
#>   tier(3): oldage_x5, oldage_x6

# Both:
kn3 <- knowledge(
  tpc_example,
  child_x1 %-->% c(child_x2, youth_x3),
  child_x2 %!-->% c(youth_x3, oldage_x5),
  tier(
    1 ~ starts_with("child"),
    2 ~ starts_with("youth"),
    3 ~ starts_with("oldage")
  )
)
kn3
#> <Knowledge: 3 tiers | 6 vars | 2 required | 2 forbidden>
#>   tier(1): child_x1, child_x2
#>   tier(2): youth_x3, youth_x4
#>   tier(3): oldage_x5, oldage_x6
#>   child_x1 %-->% child_x2 + youth_x3
#>   child_x2 %!-->% oldage_x5 + youth_x3

pc_method <- pc(engine = "bnlearn", test = "fisher_z")
disco(tpc_example, pc_method)
#> <Disco PDAG: 6 nodes | 6 edges>
#>   nodes: child_x2, child_x1, youth_x4, youth_x3, oldage_x6, oldage_x5
#>   edges: child_x1---child_x2, child_x2-->oldage_x5, child_x2---youth_x4
#>          oldage_x5-->oldage_x6, youth_x3-->oldage_x5, youth_x4-->oldage_x6
disco(tpc_example, pc_method, kn)
#> <Disco PDAG: 6 nodes | 9 edges | Knowledge: 4 required, 2 forbidden>
#> Learned graph:
#>   nodes: child_x2, child_x1, youth_x4, youth_x3, oldage_x6, oldage_x5
#>   edges: child_x1-->child_x2, child_x1-->youth_x3, child_x2---youth_x4
#>          oldage_x5-->child_x2, oldage_x5-->oldage_x6, oldage_x5-->youth_x3
#>          oldage_x6-->child_x2, oldage_x6-->youth_x3, youth_x4-->oldage_x6
#> Knowledge:
#>   vars: child_x1, child_x2, oldage_x5, oldage_x6, youth_x3, youth_x4
#>   child_x1 %-->% child_x2 + youth_x3
#>   oldage_x6 %-->% child_x2 + youth_x3
#>   child_x2 %!-->% oldage_x5 + youth_x3
disco(tpc_example, pc_method, kn2)
#> <Disco PDAG: 6 nodes | 6 edges | Knowledge: 3 tiers>
#> Learned graph:
#>   nodes: child_x2, child_x1, youth_x4, youth_x3, oldage_x6, oldage_x5
#>   edges: child_x1---child_x2, child_x2-->oldage_x5, child_x2-->youth_x4
#>          oldage_x5-->oldage_x6, youth_x3-->oldage_x5, youth_x4-->oldage_x6
#> Knowledge:
#>   tier(1): child_x1, child_x2
#>   tier(2): youth_x3, youth_x4
#>   tier(3): oldage_x5, oldage_x6
disco(tpc_example, pc_method, kn3)
#> <Disco PDAG: 6 nodes | 6 edges | Knowledge: 3 tiers, 2 required, 2 forbidden>
#> Learned graph:
#>   nodes: child_x2, child_x1, youth_x4, youth_x3, oldage_x6, oldage_x5
#>   edges: child_x1-->child_x2, child_x1-->youth_x3, child_x2-->youth_x4
#>          oldage_x5-->oldage_x6, youth_x3-->oldage_x5, youth_x4-->oldage_x6
#> Knowledge:
#>   tier(1): child_x1, child_x2
#>   tier(2): youth_x3, youth_x4
#>   tier(3): oldage_x5, oldage_x6
#>   child_x1 %-->% child_x2 + youth_x3
#>   child_x2 %!-->% oldage_x5 + youth_x3




# Large data showcase:
large_data <- as.data.frame(
  matrix(
    runif(10000),
    nrow = 100,
    ncol = 100,
    byrow = TRUE
  )
)

names(large_data) <- paste0("X_", 1:100)

kn4 <- knowledge(
  large_data,
  tier(
    seq_tiers(
      1:100,
      ends_with("_{i}")
    )
  ),
  X_1 %-->% X_2
)
kn4
#> <Knowledge: 100 tiers | 100 vars | 1 required>
#>   tier(1): X_1
#>   tier(2): X_2
#>   tier(3): X_3
#>   tier(4): X_4
#>   tier(5): X_5
#>   ... and 95 more tiers
#>   X_1 %-->% X_2

disco(large_data, pc_method, kn4)
#> <Disco PDAG: 100 nodes | 96 edges | Knowledge: 100 tiers, 1 required>
#> Learned graph:
#>   nodes: X_1, X_2, X_3, X_4, X_5, X_6, X_7, X_8, X_9, X_10, ... and 90 more
#>   edges: X_1-->X_2, X_1-->X_49, X_12-->X_36, X_13-->X_40, X_13-->X_62
#>          X_14-->X_33, X_14-->X_97, X_16-->X_80, X_17-->X_82, X_17-->X_98
#>          ... and 86 more
#> Knowledge:
#>   tier(1): X_1
#>   tier(2): X_2
#>   tier(3): X_3
#>   tier(4): X_4
#>   tier(5): X_5
#>   ... and 95 more tiers
#>   X_1 %-->% X_2

^{Created on 2026-06-19 with reprex v2.1.1}

Notes for myself:

• Deprecate summary() and old args in print/summary
• Add entry to news
• Improve vignette and examples to use new syntax of knowledge for infix operators
• Record graph class PAG since caugi doesn't support it yet. Then when printing PAGs we can say PAG instead of UNKNOWN.

[BjarkeHautop]

Whoops, sorry. Opened #49