Sra-tools do not work when installed via conda

[misialq]

For some time now, when installing sra-tools through conda (as described in the readme), none of the CLI tools actually work - all throw some mysterious certificate error.

Steps to reproduce:

1. Follow q2-fondue’s installation instructions to create a fresh env
2. After activation, just execute prefetch -v ERR1428207 and observe the error

Expected behaviour: Data for the provided ID is downloaded.

Actual behaviour: An error is thrown: The certificate is not correctly signed by the trusted CA.

Notes:

• this does not happen on Ubuntu or CentOS (only Mac: Catalina, Big Sur and Monterey)
• when using the binaries provided directly by NCBI everything seems ok
• SRA Tools' support says that it's best to use their latest versions (not available via conda) and that new versions are being only "sometimes" updated on conda
• all that means we need to (at least for now) drop the version from conda and install it manually, otherwise MacOS users run into issues

[nbokulich]

as this issue is on the end of ncbi/conda, perhaps the "fix" is to just provide instructions for installing sra-tools using the binaries released by NCBI?

I worry that this could be a persistent issue if conda will occasionally lag behind the NCBI latest release, so this might be needed as a permanent section of the documentation... (e.g., instead of temporarily pasting this in as installation instructions specifically for mac users, rather provide different alternatives for installation, and maybe also a troubleshooting/faqs section in the docs that could mention to install directly from the source if getting this certificate issue).

[misialq]

That was going to be the proposed solution. I have prepared a "universal" installation script that would install sra-tools in user's conda environment regardless of the OS. PR coming soon, just doing some final testing. I want go with a scripted solution rather than point to the original NCBI's docs as it would be better to have the toolkit installed directly in the conda environment, in case users already have some other versions installed in their system.