Update by in join block (in case of empty set intersection)

[nbenn]

Fixes #273, currently at the expense of breaking the filter block. Working on a solution to have both.

[codecov]

Codecov Report

Attention: 3 lines in your changes are missing coverage. Please review.

Comparison is base (58043a4) 74.82% compared to head (abdca1f) 77.27%.

Files	Patch %	Lines
R/block.R	62.50%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #274      +/-   ##
==========================================
+ Coverage   74.82%   77.27%   +2.44%     
==========================================
  Files          17       17              
  Lines        1903     1905       +2     
==========================================
+ Hits         1424     1472      +48     
+ Misses        479      433      -46     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[nbenn]

@DivadNojnarg I reverted a change you introduced in 1d395df, and by the commit msg you were trying to fix an issue with the filter block. I made a small change in the filter block implementation to not require this change. I GUI-checked the blocks currently in the registry and did not see any further issues. When you have updated the registry we can quickly look at whether this causes problems for the remaining blocks.

[DivadNojnarg]

There's also something about the filter expression: it is including call(+, ..., ...). I wrote this long time ago so there might be an easy way to make it nicer, that is not like this:

get("ae", envir = as.environment("package:blockr.data")) %>%
        dplyr::filter(eval(call("==", DOMAIN, "AE")))