[SkyServe] Add CLI args for resources override in sky serve {up, update}

[cblmemo]

A simplified version of #2879 . Only keep resources override.

Tested (run the relevant ones):

• Code formatting: bash format.sh
• Any manual or new tests for this PR (please specify below)

$ sky serve up examples/serve/http_server/task.yaml --cpus 4+ --gpus A100
Service from YAML spec: examples/serve/http_server/task.yaml
Service Spec:
Readiness probe method:           GET /health
Readiness initial delay seconds:  20
Replica autoscaling policy:       Fixed 2 replicas        
Each replica will use the following resources (estimated):
I 02-04 17:27:38 optimizer.py:694] == Optimizer ==
I 02-04 17:27:38 optimizer.py:705] Target: minimizing cost
I 02-04 17:27:38 optimizer.py:717] Estimated cost: $3.7 / hour
I 02-04 17:27:38 optimizer.py:717] 
I 02-04 17:27:38 optimizer.py:840] Considered resources (1 node):
I 02-04 17:27:38 optimizer.py:910] ----------------------------------------------------------------------------------------------
I 02-04 17:27:38 optimizer.py:910]  CLOUD   INSTANCE        vCPUs   Mem(GB)   ACCELERATORS   REGION/ZONE     COST ($)   CHOSEN   
I 02-04 17:27:38 optimizer.py:910] ----------------------------------------------------------------------------------------------
I 02-04 17:27:38 optimizer.py:910]  GCP     a2-highgpu-1g   12      85        A100:1         us-central1-a   3.67          ✔     
I 02-04 17:27:38 optimizer.py:910] ----------------------------------------------------------------------------------------------
I 02-04 17:27:38 optimizer.py:910] 
Launching a new service 'sky-service-879a'. Proceed? [Y/n]: ^CAborted!

$ sky serve update dummy-service-name examples/serve/http_server/task.yaml --cpus 4+ --gpu
s A100
Service from YAML spec: examples/serve/http_server/task.yaml
Service Spec:
Readiness probe method:           GET /health
Readiness initial delay seconds:  20
Replica autoscaling policy:       Fixed 2 replicas        
New replica will use the following resources (estimated):
I 02-04 17:29:07 optimizer.py:694] == Optimizer ==
I 02-04 17:29:07 optimizer.py:705] Target: minimizing cost
I 02-04 17:29:07 optimizer.py:717] Estimated cost: $3.7 / hour
I 02-04 17:29:07 optimizer.py:717] 
I 02-04 17:29:07 optimizer.py:840] Considered resources (1 node):
I 02-04 17:29:07 optimizer.py:910] ----------------------------------------------------------------------------------------------
I 02-04 17:29:07 optimizer.py:910]  CLOUD   INSTANCE        vCPUs   Mem(GB)   ACCELERATORS   REGION/ZONE     COST ($)   CHOSEN   
I 02-04 17:29:07 optimizer.py:910] ----------------------------------------------------------------------------------------------
I 02-04 17:29:07 optimizer.py:910]  GCP     a2-highgpu-1g   12      85        A100:1         us-central1-a   3.67          ✔     
I 02-04 17:29:07 optimizer.py:910] ----------------------------------------------------------------------------------------------
I 02-04 17:29:07 optimizer.py:910] 
Updating service 'dummy-service-name'. Proceed? [Y/n]: ^CAborted!

• All smoke tests: pytest tests/test_smoke.py
• Relevant individual smoke tests: pytest tests/test_smoke.py::test_fill_in_the_name
• Backward compatibility tests: bash tests/backward_comaptibility_tests.sh

[cblmemo]

@Michaelvll This is ready for another look 🫡

[Michaelvll]

Thanks for the PR @cblmemo! We should get this PR in soon. Just one comment.

[Michaelvll]

Thanks for the quick fix @cblmemo! LGTM.