docs: fpga: mgr: document parse_header() callback

success or negative error codes otherwise.

The programming sequence is::

 1. .write_init

 2. .write or .write_sg (may be called once or multiple times)

 3. .write_complete

 1. .parse_header (optional, may be called once or multiple times)

 2. .write_init

 3. .write or .write_sg (may be called once or multiple times)

 4. .write_complete

The .parse_header function will set header_size and data_size to

struct fpga_image_info. Before parse_header call, header_size is initialized

with initial_header_size. If flag skip_header of fpga_manager_ops is true,

.write function will get image buffer starting at header_size offset from the

beginning. If data_size is set, .write function will get data_size bytes of

the image buffer, otherwise .write will get data up to the end of image buffer.

This will not affect .write_sg, .write_sg will still get whole image in

sg_table form. If FPGA image is already mapped as a single contiguous buffer,

whole buffer will be passed into .parse_header. If image is in scatter-gather

form, core code will buffer up at least .initial_header_size before the first

call of .parse_header, if it is not enough, .parse_header should set desired

size into info->header_size and return -EAGAIN, then it will be called again

with greater part of image buffer on the input.

The .write_init function will prepare the FPGA to receive the image data. The

buffer passed into .write_init will be at most .initial_header_size bytes long;

buffer passed into .write_init will be at least info->header_size bytes long;

if the whole bitstream is not immediately available then the core code will

buffer up at least this much before starting.