I had one mirror that was going bad so I purchased a replacement on eBay some time ago, just never got around to mounting it. I had to make a backer plate for it and just used cyanoacrylate to bond the mirror to the steel plate.
While I was at it, I recalled a post from last year from Chris DeGerolamo
where he provided a link regarding making mirrors from old hard disk platters. The method in the instruction was to use a hole cutter to cut round "mirrors" from the platter. For my GCC machine I realized there wasn't any good reason to make the mirrors round, so I just masked a disk and cut a rectangle on a shear. (I don't know if other manufacturers require the mirror to be round.)
Anyway the cutting stresses did distort the reflected image about 2-3 mm all around the rectangle, but the incident beam would not have been in that area anyway so it was not a concern. I did a quick raster test and it looked fine. I had just finished cutting 6 mm acrylic, so I used the same settings and I was able to cut through the 6 mm fine with the aluminum mirror in place.
This is not a quantitative test by any means but it does seem like a viable way to make a mirror, at least on an emergency basis. I think it is a lot easier to cut rectangles or squares than try to make disks.
I have no idea of the reflectance of a HD mirror as opposed to a gold mirror - if you used these for all 4 locations there would probably be a performance hit. I'd be curious to know what it is, but it would take a power meter to get any quantitative data and I don't have one.