Enabling more CI runners in our pool

[nvukobratTT]

Currently, we have:

• 4xn150
• 2xn300

For better balancing, it'll be good to align those to:

• 8xn150
• 8xn300

Main reasons are:

• We're introducing Model nightly runs
• We're introducing Op Sweeps nightly runs
• We're introducing Model Analysis nightly runs
• When we have more than 3 active PR, other ones are blocked. It's common to have more than 3 as there are other folks next to our core team that contribute to FFE (~9 additional collaborators)
• We're planing to split our test runs to cover both n150 and n300 in parallel (e.g. push pipeline to run on both chip variants). This is mostly due to the inconsistencies we detected between n150 and n300.