This improves my PC's performance on RS2 Hoth by... 0.1% or so, which I
think is within the margin of error. But this change also cuts down on
boilerplate.

The TextureInfo constructor creates a vector of MipLevels. This could be
good for performance if MipLevels are accessed very often for each
TextureInfo, but that's not the case. Dolphin creates thousands of
TextureInfos per second that it never accesses the mipmap levels of
because there's a hit in the texture cache, and in the uncommon case of
a texture cache miss, the mipmap levels only get looped through once.

To make the common case of texture cache hits as fast as possible, let's
not create a vector in the TextureInfo constructor. This commit
implements a custom iterator for MipLevels instead.

In my testing on the Death Star level of Rogue Squadron 2, this speeds
up TextureInfo::FromStage by 200%, giving an overall emulation speedup
of a bit over 1%. Results on the Hoth level are even better, with
TextureInfo::FromStage being close to 300% faster and overall emulation
being over 4% faster. (Single core, no GPU texture decoding.)

Using the `-Wnrvo` flag introduced by GCC 14, I identified a few places where NRVO was clearly intended, but is fumbled.

Now only VertexLoader remains... But that one might be tricky.

To ensure memory safety, callers of GetPointer have to perform a bounds
check. But how is this bounds check supposed to be performed?
GetPointerForRange contained one implementation of a bounds check, but
it was cumbersome, and it also isn't obvious why it's correct.

To make doing the right thing easier, this commit changes GetPointer to
return a span that tells the caller how many bytes it's allowed to
access.

Gets this out of the global namespace.

m_stage is initialized last

This was added in 0b9a72a62d38481ff08f742b2318d07f29ff5dff but became irrelevant in 70f9fc4e7526fc9cfc008a43cd33229f62be99b6 as the check is now self-explanatory due to a rejiggering of the bitfields.