gpu: relax multiple-of-4 requirement on pl_buf_read

Turns out with the way this is implemented there is effectively no need
for the multiple-of-4 restriction, since the actual access to the buffer
is via the host. So drop that requirement.

Not bumping the API version because it's such a minor change in
functionality.

src/gpu.c

@@ -799,7 +799,6 @@ bool pl_buf_read(const struct pl_gpu *gpu, const struct pl_buf *buf,
 {
     require(buf->params.host_readable);
     require(buf_offset + size <= buf->params.size);
-    require(buf_offset == PL_ALIGN2(buf_offset, 4));
 
     const struct pl_gpu_fns *impl = TA_PRIV(gpu);
     return impl->buf_read(gpu, buf, buf_offset, dest, size);

src/include/libplacebo/gpu.h

@@ -576,9 +576,8 @@ void pl_buf_destroy(const struct pl_gpu *gpu, const struct pl_buf **buf);
 void pl_buf_write(const struct pl_gpu *gpu, const struct pl_buf *buf,
                   size_t buf_offset, const void *data, size_t size);
 
-// Read back the contents of a buffer, starting at a given offset (must be a
-// multiple of 4) and up to a given size, storing the data into *dest.
-// Returns whether successful.
+// Read back the contents of a buffer, starting at a given offset, storing the
+// data into *dest. Returns whether successful.
 //
 // Note: This operation will never block, so reading from a buffer that is
 // currently being written to results in the read memory regions containing