rollmean post merge improvements

[jangorecki]

Towards #2778
Follow up to PR #2961

• manual page to state "experimental" at the top
• mention ?froll or ?frollmean in the NEWS item
• strict-barrier(?) issue detected in https://gitlab.com/Rdatatable/data.table/-/jobs/134968387

 > y = frollmean(x, n, algo = "exact", adaptive = TRUE) 
  First 6 of 1 (type 'character'): [1] "Error in froll(fun = \"mean\", x = x, n = n, fill = fill, algo = algo, align = align,  : \n  cannot get data pointer of 'NULL' objects\n"
  Modes: numeric, character
  Lengths: 1000, 1
  Attributes: < target is NULL, current is list >
  target is numeric, current is try-error
  > 
  > x = c(1:4,2:5,4:6,5L)
  > n = c(2L, 2L, 2L, 5L, 4L, 5L, 1L, 1L, 2L, 3L, 6L, 3L)
  > ans1 = ama(x, n)
  > ans2 = frollmean(x, list(n), adaptive=TRUE)
  Error in froll(fun = "mean", x = x, n = n, fill = fill, algo = algo, align = align,  : 
    cannot get data pointer of 'NULL' objects