Tune workflow multiplicities for 4GPU Pb-Pb and pp async workflows

[davidrohr]

No description provided.

[github-actions]

REQUEST FOR PRODUCTION RELEASES:
To request your PR to be included in production software, please add the corresponding labels called "async-" to your PR. Add the labels directly (if you have the permissions) or add a comment of the form (note that labels are separated by a ",")

+async-label <label1>, <label2>, !<label3> ...

This will add <label1> and <label2> and removes <label3>.

The following labels are available
async-2022-pp-apass4
async-2023-pbpb-apass3
async-2023-pbpb-apass4
async-2022-pp-apass6-2023-PbPb-apass2
async-2022-pp-apass4-accepted
async-2022-pp-apass6-2023-PbPb-apass2-accepted
async-2023-pbpb-apass3-accepted
async-2023-pbpb-apass4-accepted
async-2023-pp-apass4
async-2023-pp-apass4-accepted
async-2024-pp-apass1
async-2022-pp-apass7
async-2024-pp-cpass0
async-2024-pp-cpass0-accepted
async-2022-pp-apass7-accepted
async-2024-pp-apass1-accepted

[davidrohr]

@chiarazampolli : This should improve the throughput of async Pb-Pb processing on the EPNs.
Now that ITS tracking needs less memory and sec vtx needs less CPU, we have to run more ITS trackers in parallel to fully load all CPU cores.