This is great. But I have a few suggestions.

• We need to merge the 4.x branch (I tested your PR after doing this locally)
• For the list formatter, there is a trailing semicolon that shouldn't be there
  
  I see for Tripal 3 for the list version of the formatter, the displayed value is generated by
  $locations[] = $srcfeature . ':' . $fmin . '..' . $fmax . $strand;
  and that phase is ignored. You are including phase, when present, which is good, and not including null values, also good.
  I think we just need to do like you find in tripal_chado/src/api/tripal_chado.query.api.php:928: $sql = mb_substr($sql, 0, -2); // Get rid of the trailing comma & space.
• For the table formatter, in Tripal 3 it looks like
  
  and here it looks like
  

Note that you are displaying the gene ID instead of the name of the sequence that it is located on, this needs to be changed to the name of the sequence that it is located on.
Update: This cannot currently be done, see issue #1767

• same change for list formatter for the name of the sequence

• My personal preference would be to have the "Strand" displayed before "Phase" - but that's just my preference.

• Strand should be displaying + - something is wrong here (the gene is feature_id=3 for my test)

select F.*, T.name from feature F left join cvterm T on F.type_id=T.cvterm_id;
 feature_id | dbxref_id | organism_id |             name             |                         uniquename                          | residues | seqlen |           md5checksum            | type_id | is_analysis | is_obsolete |      timeaccessioned       |      timelastmodified      |      name       
------------+-----------+-------------+------------------------------+-------------------------------------------------------------+----------+--------+----------------------------------+---------+-------------+-------------+----------------------------+----------------------------+-----------------
          3 |           |           1 | orange1.1g015632m.g          | orange1.1g015632m.g                                         |          |      0 |                                  |     165 | f           | f           | 2024-01-24 02:41:43.475274 | 2024-01-24 02:41:43.475274 | gene

select * from featureloc where feature_id=3;
featureloc_id | feature_id | srcfeature_id |  fmin   | is_fmin_partial |  fmax   | is_fmax_partial | strand | phase | residue_info | locgroup | rank 
---------------+------------+---------------+---------+-----------------+---------+-----------------+--------+-------+--------------+----------+------
             2 |          3 |             1 | 4058459 | f               | 4062210 | f               |      1 |       |              |        0 |    0

Edit: You can fix this problem with quotes

      $strand_symb = match( $strand_val ) {
        '-1' => '-',
        '1' => '+',
        default => 'unknown',
      };

• for the case of a truly unknown strand, I would prefer a blank instead of unknown but again my personal preference.

With my suggested changes I built a new docker and followed the testing steps. (I loaded the fasta, then the gff3.)
Coordinates appear as they should

Unfortunately, with the table formatter, the strand is not showing

I can add a test value for phase
sitedb=> update featureloc set phase=2 where featureloc_id=2;
and I have added another code suggestion to fix the strand, then the table format works too

Excellent, with these changes this is the exact code I tested, so I can now approve