spawn launch: should auto-detect minimum volume size from AMI snapshot

[scttfrdmn]

Summary

When launching from a custom AMI whose root snapshot is larger than spawn's default volume size, spawn launch fails with:

api error InvalidBlockDeviceMapping: Volume of size 20GB is smaller than snapshot 'snap-0e4706350e689...', expect size >= 80GB

The user is forced to pass --volume-size 80 explicitly, even though the correct minimum is already encoded in the AMI's block device mapping.

Expected behaviour

spawn launch --ami ami-0abc... should query the AMI's BlockDeviceMappings and use max(default_size, snapshot_minimum) as the volume size automatically — the same logic that the EC2 console applies when launching from a custom AMI.

Steps to reproduce

1. Bake a custom AMI with an 80 GB root volume (common for data science / bioinformatics AMIs with large pre-installed databases)
2. spawn launch my-instance --ami <custom-ami-id> --instance-type t4g.small -y
3. → InvalidBlockDeviceMapping error

Workaround

Pass --volume-size 80 explicitly. This requires users to know the AMI's snapshot size out of band.

Context

https://github.com/scttfrdmn/aws-microbiome-demo — the demo bakes an AMI with 80 GB to fit Kraken2 + MetaPhlAn databases (~40 GB). Every instance launched from this AMI needs --volume-size 80, which isn't obvious and isn't documented anywhere in the AMI metadata shown to users.

[scttfrdmn]

This is now also blocking nf-spawn (spore-host/nf-spawn#21): when nf-spawn dispatches tasks with --ami ami-xxx, spawn defaults to 20 GB and fails with InvalidBlockDeviceMapping: Volume of size 20GB is smaller than snapshot, expect size >= 40GB. If spawn auto-detected the minimum volume size from the AMI's BlockDeviceMappings, neither spawn users nor nf-spawn plugin authors would need to track this manually.