Use micromamba activate instead of mamba activate

The shell macros are only defined for micromamba
but not for mamba.

This brings consistency.

But then, what was making it working before?

Signed-off-by: Julien Jerphanion <git@jjerphan.xyz>

Co-authored-by: Sylvain Corlay <sylvain.corlay@gmail.com>
Co-authored-by: Hind Montassif <hind.montassif@gmail.com>

repo2docker/buildpacks/conda/activate-conda.sh

@@ -7,8 +7,8 @@ done
 if [[ "${KERNEL_PYTHON_PREFIX}" != "${NB_PYTHON_PREFIX}" ]]; then
     # if the kernel is a separate env, stack them
     # so both are on PATH, notebook first
-    mamba activate ${KERNEL_PYTHON_PREFIX}
-    mamba activate --stack ${NB_PYTHON_PREFIX}
+    micromamba activate ${KERNEL_PYTHON_PREFIX}
+    micromamba activate --stack ${NB_PYTHON_PREFIX}
 
     # even though it's second on $PATH
     # make sure CONDA_DEFAULT_ENV is the *kernel* env
@@ -17,5 +17,5 @@ if [[ "${KERNEL_PYTHON_PREFIX}" != "${NB_PYTHON_PREFIX}" ]]; then
     # which only contains UI when the two are different
     export CONDA_DEFAULT_ENV="${KERNEL_PYTHON_PREFIX}"
 else
-    mamba activate ${NB_PYTHON_PREFIX}
+    micromamba activate ${NB_PYTHON_PREFIX}
 fi