Function getPerioSummary

Whole-mouth periodontal summary over the active chart: total charted sites, how many bled on probing, the derived %BOP, and the single worst (deepest) CAL reading with the tooth it's on, plus the deepest raw pocket depth recorded anywhere. %BOP = bleedingSites / chartedSites — a site flagged bop without ever being charted can't exist (see setPerioSite()), so this ratio can never exceed 100%. Returns zeros/nulls (never NaN) when nothing has been charted anywhere.

maxFurcation (SP-perio P2b Task 2): the single highest Glickman grade (1-4) recorded on ANY furcation entrance anywhere in the mouth, null when nothing has been graded. Deliberately its own pass over ALL_TEETH (not folded into the pd-site loop above) since a tooth can carry furcation data independently of whether it has any charted perio site.

plaquePercent (SP-perio P2b Task 3): whole-mouth O'Leary Plaque Index — (total plaque surfaces charted across present teeth) / (present-teeth * 4) * 100, one decimal (same rounding as bopPercent above), 0 when there are no present teeth (NOT null — matches bopPercent's zero-not-null convention, unlike avgPd/avgCal). "Present" here means a REAL tooth in the mouth — toothSelection is neither "none" (missing) nor "implant" — deliberately the SAME predicate isToothPresent() uses (O'Leary is a tooth-focused index: a missing tooth has no surfaces to chart, an implant has no natural tooth surface). This is intentionally looser than perioRowHidden() (which additionally excludes under-gum/ extraction-socket teeth, since THOSE have no probeable periodontal pocket) — an under-gum or extraction-socket "tooth" still occupies a slot in the arch and can carry visible/recorded plaque on the socket/gum surface, so it still counts toward the denominator here. Deliberately its own pass over ALL_TEETH (not folded into the pd-site loop above) since a tooth can carry plaque data independently of whether it has any charted perio site, mirroring maxFurcation above. A tooth never touched at all (no entry in the active chart map) is skipped entirely, same convention maxFurcation's loop and getOdontogramSummary() use.

SP-perio PG-D Task 5 additions (own pass over ALL_TEETH, same "absence = not charted" convention as maxFurcation/plaquePercent above):

• piScore/giScore: mean of ALL charted PI/GI surface grades across the whole mouth (sum of grades / number of CHARTED surfaces, one decimal) — null when nothing is charted anywhere (mirrors avgPd/avgCal's null-when-none, NOT bopPercent/plaquePercent's zero-when-none, since a 0 mean would misleadingly read as "all surfaces charted healthy").
• kgDeficientTeeth: count of teeth with a charted kg narrower than 2mm (kg != null && kg < 2) — an uncharted tooth (kg === null) is never counted.
• gtDistribution/millerDistribution: per-value counts across the whole mouth, excluding the "not charted" skip value (unknown/ none respectively) — mirrors how maxFurcation's loop only counts graded entrances.

SP-perio PG-E Task 3 addition (own pass, same convention):

• mpiScore/mbiScore: mean of ALL charted mPI/mBI surface grades across implant teeth (sum of grades / number of CHARTED surfaces, one decimal) — null when nothing is charted anywhere, mirrors piScore/giScore exactly. mPI/mBI are implant-only axes (the setter is a no-op on a non-implant tooth), so guarding the accumulator on s.toothSelection === "implant" is belt-and-suspenders — a non-implant tooth's mpi/mbi maps can never be non-empty — but makes the implant-only intent explicit at the read site too.