fixed a typo where the wrong dpert was being used when x=0 for the relative mode

it was using eps, rather than input dpert. Fixes #14
also some CI updates

Author: jacobwilliams

.github/workflows/CI.yml

@@ -11,7 +11,7 @@ jobs:
       matrix:
         os: [ubuntu-latest]
         gcc_v: [10] # Version of GFortran we want to use.
-        python-version: [3.9]
+        python-version: [3.11]
     env:
       FC: gfortran-${{ matrix.gcc_v }}
       GCC_V: ${{ matrix.gcc_v }}
@@ -31,7 +31,7 @@ jobs:
       uses: ts-graphviz/setup-graphviz@v1
 
     - name: Setup Fortran Package Manager
-      uses: fortran-lang/setup-fpm@v5
+      uses: fortran-lang/setup-fpm@v7
       with:
         github-token: ${{ secrets.GITHUB_TOKEN }}
 
@@ -80,7 +80,8 @@ jobs:
 
     - name: Deploy Documentation
       if: github.ref == 'refs/heads/master'
-      uses: JamesIves/github-pages-deploy-action@v4.4.1
+      uses: JamesIves/github-pages-deploy-action@v4.7.3
       with:
         branch: gh-pages # The branch the action should deploy to.
         folder: doc  # The folder the action should deploy.
+        single-commit: true

src/numerical_differentiation_module.f90

@@ -2957,7 +2957,7 @@ subroutine compute_perturb_vector(me,x,dpert,perturb_mode,dx)
 
     ! make sure none are too small:
     where (dx<eps)
-        dx = eps
+        dx = dpert
     end where
 
     end subroutine compute_perturb_vector