Use clap for PGO training on Windows

Using RA takes too much memory.

Author: Kobzol

.github/workflows/release.yaml

@@ -30,10 +30,10 @@ jobs:
           - os: windows-latest
             target: x86_64-pc-windows-msvc
             code-target: win32-x64
-            pgo: true
+            pgo: clap-rs/clap
           - os: windows-latest
             target: i686-pc-windows-msvc
-            pgo: true
+            pgo: clap-rs/clap
           - os: windows-latest
             target: aarch64-pc-windows-msvc
             code-target: win32-arm64
@@ -91,11 +91,11 @@ jobs:
 
       - name: Dist (plain)
         if: ${{ !matrix.zig_target }}
-        run: cargo xtask dist --client-patch-version ${{ github.run_number }} ${{ matrix.pgo && '--pgo' || ''}}
+        run: cargo xtask dist --client-patch-version ${{ github.run_number }} ${{ matrix.pgo && format('--pgo {0}', matrix.pgo) || ''}}
 
       - name: Dist (using zigbuild)
         if: ${{ matrix.zig_target }}
-        run: RA_TARGET=${{ matrix.zig_target}} cargo xtask dist --client-patch-version ${{ github.run_number }} --zig ${{ matrix.pgo && '--pgo' || ''}}
+        run: RA_TARGET=${{ matrix.zig_target}} cargo xtask dist --client-patch-version ${{ github.run_number }} --zig ${{ matrix.pgo && format('--pgo {0}', matrix.pgo) || ''}}
 
       - run: npm ci
         working-directory: editors/code