Buffer.sync and Buffer.syncFromArrayLists computed the new buffer size
in bytes (req_bytes * 2) and then multiplied by @sizeOf(T) again when
passing it to newBufferWithLength:, allocating data.len * sizeOf(T)^2 * 2
bytes. For the 32-byte CellText buffers this is a 64x over-allocation
and for the 4-byte CellBg buffers 8x, per swap-chain frame (e.g. ~9.4MB
instead of ~300KB per frame for a full 120x40 screen of text).
Match the OpenGL buffer implementation: track the new length in units
of T and multiply by @sizeOf(T) exactly once.
This commit is very large, representing about a month of work with many
interdependent changes that don't separate cleanly in to atomic commits.
The main change here is unifying the renderer logic to a single generic
renderer, implemented on top of an abstraction layer over OpenGL/Metal.
I'll write a more complete summary of the changes in the description of
the PR.
Discrete GPUs cannot use the "shared" storage mode. This causes
undefined behavior right now, and I believe it's what's causing a
problem on Intel systems with discrete GPUs with "inverted" cells.
This commit also sets the CPU cache mode to "write combined" for our
resources since we don't read them back so Metal can optimize them
further with this hint.
This will allow for unlimited glyphs per row, eliminating the issue run in to with multi-substitution glyphs and combining characters which can result in more glyphs in a row than there are columns.
