Local development improvements

[BioGeek]

This PR contains a few improvements which make local development and running the benchmark easier.

• When developing and testing your algorithm (and those of others) locally, you often want to run the benchmark only for a single algorithm and a single dataset. This is now possible, for example:
  ./run.sh sample_data/9_species_human pi-HelixNovo
• If you want to run the whole benchmark of all algorithms on all (sample) datasets, you can now do make run_all.
• The container_template.def recommends to copy the whole algorithms/algorithm_name folder to /algo. So the first time you build your image, a container.sif file will be created in the algorithms/algorithm_name folder. But the second time you build your image, when build_apptainer_images.sh asks if you want to force rebuild your image, it will copy the existing container.sif into the new container.sif. And so on each time you rebuild your image, just like a Matryoshka doll. I noticed this issue when the size of my container.sif had ballooned to 64GB. As a short term solution, I have added a make clean command that removes all container images and a make build_apptainer command that force builds all apptainer images. But a better long term solution is to adapt all the container.def files to only copy the files that are needed into the /algo folder.
• I've replaced the instances where there were several subsequent pip install <package> lines one after the other with a single pip install <list of packages>. If pip can see all the packages at once it can resolve conflicts between package dependencies more effectively. It also avoids redundant resolution steps, making the installation process faster.
• The flag option in run.sh to recalculate all algorithm outputs didn't work for me, so I've set the default value to true.