data_precision_fmt for precision based format

[gmbecker]

E.g., mean(x) should have 1 decimal points more effective precision than x did (only really makes sense when x was measured to finite, relatively low precision)

Very rough, probably wrong mockup of behavior

data_precision_digs <- function(x, digits_more= 1) {
  stopifnot(is.numeric(x)) 
  chr <- as.character(x)
  chr <- gsub("e[-+][0-9]$", "", chr)
  ## leading zeros (without a .) and trailing zeros (ever) are not precision
  chr <- gsub("[0]*([.123456789][[:digit:].]*[123456789])[0.]*$", "\\1", chr) 
  chr <- gsub(".", "", chr, fixed = TRUE)
  max(nchar(chr)) + digits_more
}

data_precision_fmt <- function(x, digs_more = 1) {
   digs <- data_precision_digs(x, digs_more)
   digs_to_fmt(digs)
}

digs_to_fmt <- function(digs) {
## awesome code goes here
}

[edelarua]

Feature implemented in tern: insightsengineering/tern#1051