Regression in staging!  Prod works better (even with ?env=staging to ensure seeing same quotas). However, I suspect prod computation is not 100% right either.

Seen in OSD Wizard, haven't tested ROSA yet — but listing both in card title to make sure fix gets tested in both.

Steps to reproduce

(As any quota problem, this depends on exact quota... This is one scenario that reproduces with current UI team's staging org 1wuVGGV6SCmD8ya6yRGEJzvmVuC)

1. Open Create cluster -> OSD cluster wizard
2. Choose Annual billing, Red Hat cloud account
3. Choose AWS
4. Choose single AZ
5. Choose large "Compute node instance type". Specifically, I'm seeing this for largest Memory Optimized — r5.4xlarge, and for both Compute Optimized — c5.2xlarge, c5.4xlarge.

Actual behavior

"Compute node count" dropdown becomes disabled with "You do not have enough quota for this option. Contact sales to purchase additional quota." tooltip.
This is inconsistent!

Expected behavior

• MachineTypeSelection / NodeCountInput should agree on which types are possible.
  Either an instance type should not be listed, or there should be some valid number of nodes.
• The maximum number of nodes listed by NodeCountInput should match the maximum the backend would accept.
  (fastest way to test that is ocm create cluster -dry-run ... playing with -compute-nodes=... flag)

Note that even for smaller instance types, where NodeCountInput is enabled, I'm seeing it offer different maximum between staging UI & prod UI ?env=staging (to ensure both getting same quota), so it's important to confirm backend's idea of maximum.