Change: skip realsprite palette validation step iff output is nfo AND…

[andythenorth]

… a palette is specified as run-time parameter

Saves ~10% of compile time for me by skipping redundant palette check when using nfo output (grfcodec will validate the palettes later).

[andythenorth]

Diff looks wrong currently with respect to black formatting.

[Eddi-z]

isn't this logic backwards? if i give a palette as parameter, i'd expect that the palette gets checked, otherwise i don't care?

also, does grfcodec do the same kind of checks on the palette as nmlc?

[andythenorth]

isn't this logic backwards? if i give a palette as parameter, i'd expect that the palette gets checked, otherwise i don't care?

Fail safe.
Vanishingly few nml users will ever need or want to disable that check. Most will not understand it as an optional parameter (or find it).

EDIT: no you're right the logic does appear to be backwards in the current implementation. It's an unfortunate artefact of needing to know explicitly which palette to set for action 14, which is otherwise currently inferred during the palette check (the approach to inferring palette seems extremely odd to me, but eh, maybe I missed something).

also, does grfcodec do the same kind of checks on the palette as nmlc?

No. It checks for invalid white pixels and invalid colour profiles. It does not validate, e.g. animated pixels etc. Accepted risk.

[andythenorth]

Superceded by #322 - this route involves inferring that we want to skip validation if we're outputting .nfo, which is unwise.