Allign tcl/target/gd32vf103.cfg with the mainline

[en-sc]

There were two patches to RISC-V OpenOCD that caused the conflicts that were misresolved:

1. Improve support for GD32VF103 MCU #538, that was further upstreamed at https://review.openocd.org/c/openocd/+/6957.
2. Update gd32vf103.cfg #640

I've reverted #640 -- Specifying the adapter speed seems better placed in board config or in the command line and transport select jtag is already present in the mainline version.
After that the file was reset to the version for upstream (7ac389c).

[en-sc]

@groencode, @tchebb, FIY.

[en-sc]

@tom-van, could you please take a look at this one?

[tom-van]

We have a couple of old pending patches where author did not respond to my comments
6959: tcl/target/gd32vf103: copy a few minor settings from riscv-openocd | https://review.openocd.org/c/openocd/+/6959

• adding $_TARGETNAME riscv set_enable_virt2phys off is undoubtedly good, but the command name changed in riscv sync series
• adding bs tap just prevents some messages at startup/reconnect and surely cannot be combined with reset_config srst_nogate

6958: tcl/target/gd32vf103: add main flash alias | https://review.openocd.org/c/openocd/+/6958

• probably nonsense

Yes, merge it as-is and later we can add riscv virt2phys_mode off in both upstream and here.

[tchebb]

I'm the author of the upstream patches as well as the more fiddly bits of the downstream version. Sorry for disappearing: I was busy when the patches were reviewed and then had forgotten a lot of the details once I had time again.

I'll check out the current state of things later today and respond to the upstream patches as appropriate.

[tom-van]

Great, welcome back! I still have the Sipeed Longan Nano somewhere on a shelf

[tchebb]

I've rebased and responded to both outstanding upstream changelists. I recall now that it was the changes you made to https://review.openocd.org/c/openocd/+/6957 that I kept getting stuck on, not the simpler feedback in the other two.

+1 to this PR, too. gd32vf103.cfg seems to have gotten quite mangled in the merges, so this will much improve it even if you leave out the pending upstream CLs.