Fix conda environment and simplify installation and setup through git submodule

[richardstoeckl]

Hello, I hope you are doing well.

this PR does two things regarding the installation and setup process, therefore I didn't want to separate them into two PRs.

1. Fix and simplify the conda environment

Some dependencies and packages are not available in their specified version through the conda-forge or bioconda channels anymore (i.e. tensorflow). While some would still available through the anaconda channel, this has some restrictive licensing terms (see the Anaconda Terms of Service). Also, the environment file specified several channels that are either unnecessary (axfeh) or outdated (biobuilds).

Changes:

• Moved tensorflow, tensorboard, and keras to pip installation, keeping the same versions.
• Relaxed the Perl version constraint to allow the perl-compress-bgzf module build.
• Removed axfeh::spacerplacer (leftover from another project). If it is still needed, the bioconda::spacerplacer package should be used instead.
• Removed unnecessary channels

2. Automate CRISPRcasIdentifier setup

CRISPRidentify previously required users to manually download and place CRISPRcasIdentifier v1.1.0 in the correct directory. Additionally, models had to be downloaded from Google Drive and placed in the correct location manually. This PR automates this process by adding CRISPRcasIdentifier as a git submodule, and using gdown to download the correct models. The manual process remains possible without changes.

Changes:

• Introduces BackofenLab/CRISPRcasIdentifier as a git submodule pinned to v1.1.0. With git clone --recurse-submodules, the repository is now included automatically in the expected location.
• Adds gdown as a dependency and includes README.md commands to download the required models into the correct paths, following the procedure documented in BackofenLab/CRISPRcasIdentifier.

---

I cloned CRISPRidentify fresh with --recurse-submodules and followed the instructions using gdown for the model files. I then rebuilt the environment from scratch and confirmed that python CRISPRidentify.py --input_folder TestInput runs end-to-end.

I am open to feedback, and hope to see you soon again at another conference.

Best,
Richard