Fix namespace lookup failures in superbData without attaching package

[achetverikov]

Summary

This PR fixes a namespace lookup failure in superbData() / superbPlot() when users pass namespaced strings like statistic = "superb::mean" and errorbar = "superb::CI" without attaching superb to the search path.

It also removes global-environment side effects in superbPlot() by eliminating dynamic wrapper creation (superbSTATISTIC / superbERRORBAR in .GlobalEnv).

Problem

1) HyunhFeldtEpsilon lookup failure (within-subject CM/LM path)

In superbPlot(), diagnostics used:

• plyr::ddply(..., .fun = "HyunhFeldtEpsilon", ...)
• plyr::ddply(..., .fun = "WinerCompoundSymmetryTest", ...)
• plyr::ddply(..., .fun = "MauchlySphericityTest", ...)

With string .fun, plyr resolves via match.fun() and caller/search-path lookup.
When superb is not attached, this can fail with:

object 'HyunhFeldtEpsilon' of mode 'function' was not found

2) Global environment mutation

superbPlot() created temporary wrappers in .GlobalEnv via eval(parse(...), envir = globalenv()):

• superbSTATISTIC
• init.superbSTATISTIC
• superbERRORBAR

This introduces side effects and can interfere with package usage patterns.

Changes

Commit 1: Use function objects in ddply sphericity diagnostics

In R/superbPlot.R, replaced string .fun with function objects:

• "HyunhFeldtEpsilon" -> HyunhFeldtEpsilon
• "WinerCompoundSymmetryTest" -> WinerCompoundSymmetryTest
• "MauchlySphericityTest" -> MauchlySphericityTest

This removes fragile search-path function lookup in the CM/LM diagnostics block.

Commit 2: Remove globalenv wrappers for statistic and errorbar

In R/superbPlot.R:

• Removed eval(parse(...), envir = globalenv()) wrapper generation.
• Resolved statistic/errorbar/init functions as local function objects.
• Replaced string-based do.call("...") calls with direct function-object calls.
• Kept existing user API (statistic/errorbar as strings, including "namespace::name").
• Preserved existing feedback messages and behavior.

Repro (before)

data(sleep, package = "datasets")
sleep_wide <- reshape(sleep, direction = "wide", idvar = "ID", timevar = "group")

superb::superbData(
  sleep_wide,
  WSFactors   = "group(2)",
  variables   = c("extra.1", "extra.2"),
  adjustments = list(purpose = "single", decorrelation = "CM"),
  statistic   = "superb::mean",
  errorbar    = "superb::CI"
)

Could fail with:

object 'HyunhFeldtEpsilon' of mode 'function' was not found

Validation

• Full test suite run: PASS 291, FAIL 0, WARN 0, SKIP 0.
• Side-by-side output comparison (CRAN superb vs patched) across between/within/mixed examples, with and without gamma:
  • Outputs are identical (max_abs_diff = 0) for all previously working cases.
  • Only behavior change: previously failing non-attached within-subject case now works.
• Confirmed no global env artifacts after calls:
  • exists("superbSTATISTIC", .GlobalEnv, inherits = FALSE) -> FALSE
  • exists("superbERRORBAR", .GlobalEnv, inherits = FALSE) -> FALSE