Goal

As I user, I want to automatically configure virtual disks for my VMs with block limits that match the host storage so that performance isn't negatively impacted by a mismatch.

One way to achieve this is adding an "auto" value to the existing block limit properties, e.g. -device virtio-blk,discard_granularity=auto. This "auto" value should then be added for the following properties:

• logical_block_size (matches backend request_alignment, but at least 512)
• opt_io_size (matches backend opt_transfer)
• discard_granularity (matches backend pdiscard_alignment)

Covering physical_block_size and min_io_size as well would be nice, but this is not a value QEMU detects for BlockLimits yet, so this would have to be a separate story. (Though there is a very limited bdrv_probe_blocksizes() that could possibly be extended to achieve this.)

Possibly the existing backend-defaults property can help, too, but in any case the interactions between it and any other changes to properties need to be considered.

Acceptance criteria

TBD