Compilers and libraries from conda instead of module load

[YuuuXie]

This PR allows to installing flare without loading any modules, thus avoid the case where users have different module version on different clusters that might cause various compatibility issues.

Three small changes are made in this PR:

• Add python 3.8 to the cmake file
• Add Conda's include folder into the search directories in cmake, such that the header files (e.g. mkl.h) installed by conda can be found.
• Increase numpy version upper bound to 1.22 (1.23 has compatibility issue with numba)

---

The compiler tools and libraries required by the flare c++ code are installed by conda, instead of loading from modules. The new installation recipe is as below:

1. Create a new conda env with python<=3.8

conda create --name flare python=3.8
conda activate flare

2. Install compilers and libraries

   • Option 1: If you want to install flare with mkl

   conda install -y gcc gxx cmake mkl-devel mkl-service openmp -c conda-forge

   • Option 2: If you want to install flare with openblas + lapacke

   conda install -y gcc gxx cmake openmp liblapacke openblas -c conda-forge

3. Install flare

git clone -b development https://github.com/mir-group/flare.git
cd flare
pip install .