speculative : fix n_outputs_max and remove draft-simple auto-enable (#23988)

* speculative : add common_speculative_n_max helper function

Extract the speculative max-draft-size logic from server_n_outputs_max
into a reusable common_speculative_n_max() function in common/speculative.

Assisted-by: llama.cpp:local pi

* cont : draft context always has n_parallel outputs

* llama : log n_outputs_max

* speculative : remove draft-simple auto-enable

* ci : enable server tests on PRs

.github/workflows/server.yml

@@ -102,7 +102,6 @@ jobs:
 
       - name: Tests
         id: server_integration_tests
-        if: ${{ !github.event.pull_request }}
         run: |
           cd tools/server/tests
           pytest -v -x -m "not slow"
@@ -116,7 +115,6 @@ jobs:
 
       - name: Tests (Backend sampling)
         id: server_integration_tests_backend_sampling
-        if: ${{ !github.event.pull_request }}
         run: |
           cd tools/server/tests
           export LLAMA_ARG_BACKEND_SAMPLING=1
@@ -169,7 +167,6 @@ jobs:
 
       - name: Tests
         id: server_integration_tests
-        if: ${{ !github.event.pull_request }}
         run: |
           cd tools/server/tests
           $env:PYTHONIOENCODING = ":replace"