add selectable (but static) SFP Base-X support - #18

One comment on the two patch files: they have a git format-patch header on a
quilt body (Index: / ==== / .orig paths instead of diff --git a/… b/…).
It applies, but git am won't take it, the newly created files lose their
new file mode, and there's no Signed-off-by.

More importantly, I think that's how a hunk went missing. 0008's diffstat says:

  .../cpu/armv8/fsl-layerscape/ls1046a_serdes.c |   2 +
  board/freescale/gateway_dk/gateway_dk.c       | 194 ++++++++++++++++++
  2 files changed, 196 insertions(+)

but the body only contains gateway_dk.c — the ls1046a_serdes.c change
(the 0x1333 table entry, I assume?) isn't in there. Header from the original
commit, body from a later refresh.

Regenerating both patches with git format-patch against a tree at
NXP_LF_SRCREV_UBOOT / NXP_LF_SRCREV_RCW should fix the format and bring
the missing hunk back in one go.

More importantly, I think that's how a hunk went missing. [...] the ls1046a_serdes.c change (the 0x1333 table entry, I assume?) isn't in there.

After creating the initial PR, I have been using quilt to manage the patchset. It hadn't occurred to me to check the generated patch header. The change in ls1046a_serdes.c was actually the addition of an 0x3133 entry, which I removed at a later point because it's not fully correct that way. 0x1333 exists out of the box in the NXP upstream sources. So, there is no hunk missing, the metadata was just out of date.

I'll re-generate both patches with git am to make sure the metadata is up to date, since quilt doesn't touch that.