Auto decide which disk to use if a disk is not specified in hardware.csv

**What would you like to be added**: Auto decide which disk to use for Bare Metal node

**Why is this needed**: Users have to figure out the disk name of each nodes to be able to deploy the cluster. With this feature, the user can skip this manual process.

Comment: it would be more feasible to modify Tinkerbell “write-disk” action to be able to auto find a disk, comparing to get the disk info from BMC. Because we need to support the use case of [no bmc](https://github.com/aws/eks-anywhere/blob/main/pkg/providers/tinkerbell/hardware/testdata/hardware_no_bmc_headers.csv). [templates rendering](https://github.com/tinkerbell/cluster-api-provider-tinkerbell/blob/main/internal/templates/templates.go#L76) process may need to be changed accordingly.

Review the Tinkerbell write-disk action context, pkg/providers/tinkerbell/hardware/testdata/hardware_no_bmc_headers.csv, and the templates rendering process in internal/templates/templates.go around line 76. Determine how disk selection should work when hardware.csv omits a disk, including the no-BMC case. Done means a Bare Metal node can deploy without manually specifying each disk.

Written by the indexing model from the issue text.