Always validate existing database and extension

[erimatnor]

This change ensures the database and extension is validated whenever
these objects aren't created, instead of only doing validation when
bootstrap=>false is passed when adding a data node.

This fixes a corner case where a data node could be added and removed
several times, even though the data node's database was already marked
as having been part of a multi-node setup.

[codecov]

Codecov Report

Merging #2772 (fce26f6) into master (71633c3) will increase coverage by 0.10%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #2772      +/-   ##
==========================================
+ Coverage   90.11%   90.22%   +0.10%     
==========================================
  Files         212      212              
  Lines       34665    34617      -48     
==========================================
- Hits        31239    31232       -7     
+ Misses       3426     3385      -41     

Impacted Files	Coverage Δ
tsl/src/data_node.c	96.08% <100.00%> (-0.18%)	⬇️
src/loader/bgw_message_queue.c	84.51% <0.00%> (-3.23%)	⬇️
tsl/src/dist_util.c	91.94% <0.00%> (-1.35%)	⬇️
tsl/src/data_node_dispatch.c	97.02% <0.00%> (-0.25%)	⬇️
src/import/planner.c	70.30% <0.00%> (+11.12%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 71633c3...fce26f6. Read the comment docs.

[k-rus]

I suggest to merge the commits into one, since the test commit checks that the non-bootrstrap option works and will fail without the first commit. I hope I haven't missed anything in the logic.

[erimatnor]

I suggest to merge the commits into one, since the test commit checks that the non-bootrstrap option works and will fail without the first commit. I hope I haven't missed anything in the logic.

All tests pass in both commits, independently. The second commit adds a test case that covers the case that didn't work as expected. Although I can certainly squash this, I personally don't see any issues with separating changes into logical units.

[k-rus]

I suggest to merge the commits into one, since the test commit checks that the non-bootrstrap option works and will fail without the first commit. I hope I haven't missed anything in the logic.

All tests pass in both commits, independently. The second commit adds a test case that covers the case that didn't work as expected. Although I can certainly squash this, I personally don't see any issues with separating changes into logical units.

The first commit implements validation of database and extension if bootstrap is false. However, I see no tests in the this commit, which checks this. It is good practice to have test in the same commit as implementation for several reasons:

1. It helps to understand the commit better.
2. It allows to cherry pick or rollback the implementation with the regression test together.

If I understood the second commit correctly, it includes a test for the first commit. Correct me if I am wrong and test is missing.