fix: software mentions counts on homepage

[dmijatovic]

Improve mention count on homepage

Closes #1326

Changes proposed in this pull request:

• Improve  homepage_counts RPC to use mentions_by_software RPC for software mentions count
• We count unique combination of mention and software id's. In other words, if multiple software packages (registered in RSD) are referenced from one paper we count this reference multiple times (for each software registered in RSD).

How to test:

• Easier to validate is to start without test data. docker compose build && docker compose up
• Login and create 2 software packages and 1 project
• In the first software package in reference papers use this DOI 10.1186/s13321-017-0220-4
• In the second software package in reference papers use this DOI 10.5194/gmd-7-267-2014
• In the project output use this DOI 10.5194/gmd-10-3167-2017
• Let the scrapers run and collect the citations (wait 15 minutes or more).
• Navigate to home page and check homepage stats. It should show 555 software mentions
  • 10.1186/s13321-017-0220-4 produces 461 software mentions
  • 10.5194/gmd-7-267-2014 produces 16 software mentions
  • 10.1016/j.softx.2020.100549 produces 14 software mentions
  • 10.1016/j.future.2018.08.004 produces 64 software mentions

Example software overview

PR Checklist:

• Increase version numbers in docker-compose.yml
• Link to a GitHub issue
• Update documentation
• Tests

[jmaassen]

Works as expected for the test cases described above.

I also tested manually adding a mention to software1 that was also found automatically. This isn't counted twice, which is correct.

When also adding this same mention manually to software2 it is counted. This is also correct.

However, if I use the same reference paper for both software 1 and software 2, the citations are only counted once. I don't think this is correct? I agree this is an artificial example though. A more realistic one is the following:

AMBER has 10.1016/j.softx.2020.100549 as one of it's reference papers. This is cited 13 times.

KernelTuner has 10.1016/j.future.2018.08.004 as one of it reference papers. This is cited 64 times.

Because these are different pieces of software, the total citation count should be 13+64 = 77. However, the RSD only reports 75, because some citations are coming from the same source. For example this book chapter 10.1007/978-3-031-69577-3_7 cites both AMBER and KernelTuner. It should count as 2, but is now only counted as 1.

[ewan-escience]

Works well, I'll leave it up to @jmaassen to decide if the semantics should be changed.

[dmijatovic]

@jmaassen I have adjusted RPC to count on unique combination of mention and software id's. Using this approach we count one paper multiple times if it references more than one RSD software entry. Can you test it again, please?

[sonarqubecloud]

Quality Gate passed for 'rsd-database'

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarCloud

[jmaassen]

Works as expected. The overall count on the landing pages makes sense now.