Treat every MK2 board as an Elite, whatever revision it reports (#819)

[enver-haase]

Closes #819.

isEliteBoard() answers from a revision table that has no entry for 0x10 — the TI HDMI board that became the C64U — so a board reporting it is told it is not an Elite. On the MK2 that is always wrong: nothing this function gates is revision dependent there. It hides the SID socket shunts and the joystick swapper, and both are present on every MK2 board.

@GideonZ's call in the issue: "The if U64 == 2 return true sounds like the best thing to do." That is what this does. The MK1 branch is the same code as before, moved under #else and not otherwise touched, so a U64 == 1 build decides exactly as it did — including the 0x14 case that reads C64_PLD_JOYCTRL to tell the two variants apart.

Not measured on hardware, and why

The defect needs a board reporting 0x10. Mine reports 0x17, for which isEliteBoard() already answered true, so there is nothing on my machine that changes. I offered in the issue to prove this red then green by forcing the revision; having built it, I think that offer was worth less than it sounded. A forced revision proves the branch, not the device, and I would rather say so than dress it up as a hardware result.

If somebody has a 0x10 board, the check is one line in the menu: the joystick swapper is selectable after this change and greyed out before it.

A second thing, while this file is open

On the MK2 isEliteBoard() gates only config items, through cfg->disable(). That sets setEnabled(false) — and the REST layer never reads that flag, neither in the listing nor on write. route_configs.cc iterates st->items and renders every one of them.

So on a 0x10 board the joystick swapper would have been greyed out in the menu and still readable and settable over PUT /v1/configs/.... That is independent of this change and survives it. Whether disable() is meant to reach the API is yours to say; I have not touched it here.

Built

make u64ii_no_esp in ghcr.io/gideonz/riscv, from a tree with every output/ and result/ under target/ removed first: 461 files, exit 0, no warning from product.cc.

make u64 and make u64_no_esp cannot run in that image at all: the MK1 BSP wants nios2-elf-gcc, which is not in it — the self-hosted runner supplies it by mounting /opt/build-tools. Unmodified master fails at the same file with the same error, so this is the image and not the change. CI builds u64_no_esp, so the MK1 path is covered by this PR's own run rather than by anyone's promise.