feat: unify AI cover generation to consistent placeholder style

Commit notes

- Replace 3 distinct styles (abstract, photographic, illustrated) with a unified minimalist placeholder aesthetic - All generated covers now share the same base style but vary through: - Variation emphasis (rounded/angular/flowing shapes) - Mood-based color palettes - Category-inspired visual elements - Pack name thematic hints - Update frontend labels from style names to variation descriptions - Update UI text to reflect 'placeholder inspiration' purpose - Maintain backwards compatibility with existing API contracts

The unified style creates a consistent, recognizable aesthetic across all generated covers while still allowing contextual variation based on pack content. This makes the covers serve as clear placeholders and inspiration rather than final artwork.

Co-authored-by: armin.naimi <[email redacted]>