find the colour table relative to the mod, not the cwd

ShinyPalette loaded data/shiny_colors.lua by guessing cwd-relative paths
when V.data was absent, which is the headless case. Run from the project
root, none of the guesses hit -- so the extraction built all 151 species,
recoloured none of them, and reported PASS. The packs were correct; there
simply were no shiny variants in them, and nothing in the output said so.

Two fixes, because either alone leaves the trap open:

  V.path is now consulted first, so the file is found relative to the MOD
  the way every other resource is. The extract test's stub sets it to
  whatever --mod it was given.

  A count of zero recoloured is now a failure. That is precisely what a
  missing colour table looks like from the outside, and a test that passes
  on it is not testing the feature. It was caught by noticing the number
  change between two runs that differed only in where they were started
  from -- which is too thin a thread to hang it on.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

lib/ShinyPalette.lua

@@ -168,8 +168,16 @@ local function loadColors()
     local ok, t = pcall(V.data, "shiny_colors")
     if ok and type(t) == "table" then colors = t; return colors end
   end
-  local tries = { "data/shiny_colors.lua",
+  -- Off disk, RELATIVE TO THE MOD rather than to the working directory.
-                  "mods/DramaticShapeVoxelMod/data/shiny_colors.lua" }
+  -- V.path is the mod's own directory (main.lua sets it; the headless
+  -- harnesses set it to whatever --mod they were given). Guessing from the
+  -- cwd instead is what made this silently find nothing when the extraction
+  -- test was run from the project root rather than from the mod: every
+  -- species built, none recoloured, and a PASS at the end of it.
+  local tries = {}
+  if V and V.path then tries[#tries + 1] = V.path .. "/data/shiny_colors.lua" end
+  tries[#tries + 1] = "data/shiny_colors.lua"
+  tries[#tries + 1] = "mods/DramaticShapeVoxelMod/data/shiny_colors.lua"
   for _, p in ipairs(tries) do
     local chunk = loadfile(p)
     if chunk then
@@ -181,6 +189,13 @@ local function loadColors()
   return nil
 end
 
+-- Whether the colour table was found at all. The extraction asks so it can
+-- say "no colours" once and loudly, rather than reporting 151 successful
+-- builds with no shiny variant among them.
+function ShinyPalette.haveColors()
+  return loadColors() ~= nil
+end
+
 -- The spec for one dex number, or nil if we have nothing for it.
 function ShinyPalette.forDex(dex)
   local all = loadColors()

tests/stadium_extract_test.lua

@@ -51,6 +51,9 @@ function V.require(name)
   return loaded[name]
 end
 V.mod = { log = { warn = function() end, info = function() end } }
+-- the mod's own directory, so a module that loads a data file finds it
+-- relative to the MOD rather than to wherever this was run from
+V.path = MOD
 
 local StadiumRom = V.require("StadiumRom")
 local StadiumBuild = V.require("StadiumBuild")
@@ -150,6 +153,16 @@ io.write(("\n%d checked, %d identical, %d differ, %d oracle files missing, "
 io.write(("shiny: %d recoloured, %d without a variant, %d malformed\n")
          :format(shinyOk, shinyMissing, shinyBad))
 
+-- NONE recoloured is a failure, not a quiet zero. It is what a missing or
+-- unfindable data/shiny_colors.lua looks like, and the first version of this
+-- test reported PASS through exactly that: 151 species built, every one of
+-- them without a shiny variant, and nothing in the output that read as
+-- wrong. A count of zero is now as loud as a malformed pack.
+if checked > 0 and shinyOk == 0 then
+  io.write("NO SPECIES RECOLOURED -- data/shiny_colors.lua was not found\n")
+  shinyBad = shinyBad + 1
+end
+
 -- The oracle diff is the load-bearing assertion and is unchanged: the shiny
 -- pass must not have moved a single byte of the normal packs. The shiny
 -- counters are additional, and a malformed variant fails the run -- a pack