216. Parse ines rom

add emulator/cartridge/ines.py::parse_ines_rom.

Why this step exists

This function turns header claims into clean PRG/CHR slices so later Cartridge and mapper code never needs to understand iNES headers or trainer offsets. It uses all of the format constants and parser models from lessons 212-215.

Suggested implementation

def parse_ines_rom(data: bytes) -> INesRom:
    header = parse_ines_header(data)
    prg_size = header.prg_rom_banks * PRG_ROM_BANK_SIZE
    chr_size = header.chr_rom_banks * CHR_ROM_BANK_SIZE
    prg_start = INES_HEADER_SIZE + (
        TRAINER_SIZE if header.has_trainer else 0
    )
    prg_end = prg_start + prg_size
    chr_start = prg_end
    chr_end = chr_start + chr_size
    if len(data) < chr_end:
        raise ValueError(
            "iNES data is too short for declared PRG/CHR ROM"
        )
    return INesRom(
        header=header,
        prg_rom=data[prg_start:prg_end],
        chr_rom=data[chr_start:chr_end],
    )

Invariants: sizes come from bank counts; PRG starts after the header and optional 512-byte trainer; CHR immediately follows PRG; and all declared bytes must exist before slicing. Do not rely on forgiving short Python slices, which would accept a truncated image silently.

Out of scope for this step

1. Extra trailing-byte policy and CHR RAM allocation are not introduced here.
2. Lesson 217 constructs a `Cartridge` from this parser result.
3. Lessons 218-220 add mapper behavior and mapper selection.