BUG #17691: Unexpected behaviour using ts_headline()

The following bug has been logged on the website:

Bug reference:      17691
Logged by:          Sebastian Patino-Lang
Email address:      sebastian.patino-lang@posteo.net
PostgreSQL version: 13.9
Operating system:   x86_64-apple-darwin19.6.0
Description:

I experience unexpected behaviour when using ts_headline() in general, but
especially when changing  MaxFragments. Given the data in
ts_headline_report.sql [1]

SELECT id,
ts_headline('english', "texts"."fulltext", to_tsquery('english', 'amazon &
world'), 'StartSel=<<, StopSel=>>') AS "preview"
FROM texts;

id=1: Highlight word is the first one in the result. Expectation: highlight
word is somewhere in the middle.
id=2: No highlight word at all.
id=3: Highlight words are the first and last one in the result. Not ideal
but ok-ish.

SELECT id,
ts_headline('english', "texts"."fulltext", to_tsquery('english', 'amazon &
world'), 'MaxFragments=1, StartSel=<<, StopSel=>>') AS "preview"
FROM texts;

id=1: Highlight word is now in the middle of the result. This is ok.
id=2: No highlight word at all.
id=3: Highlight words are in the middle part of the result. This is ok.

SELECT id,
ts_headline('english', "texts"."fulltext", to_tsquery('english', 'amazon &
world'), 'MaxFragments=2, StartSel=<<, StopSel=>>') AS "preview"
FROM texts;

id=1: Correct number of fragments (2) with highlight words are returned. 
id=2: No highlight word at all.
id=3: Correct number of fragments (2) with highlight words are returned. 

SELECT id,
ts_headline('english', "texts"."fulltext", to_tsquery('english', 'amazon &
world'), 'MaxFragments=3, StartSel=<<, StopSel=>>') AS "preview"
FROM texts;

id=1: Wrong number of fragments (2) with highlight words are returned. 
id=2: No highlight word at all.
id=3: Correct number of fragments (3) with highlight words are returned.

SELECT id,
ts_headline('english', "texts"."fulltext", to_tsquery('english', 'amazon &
world'), 'MaxFragments=4, StartSel=<<, StopSel=>>') AS "preview"
FROM texts;

id=1: Wrong number of fragments (2) with highlight words are returned. 
id=2: No highlight word at all.
id=3: Correct number of fragments (4) with highlight words are returned.

... and so on. Until MaxFragments=6 where for id=1 suddenly more fragments
(4) get returned.

SELECT id,
ts_headline('english', "texts"."fulltext", to_tsquery('english', 'amazon &
world'), 'MaxFragments=4, StartSel=<<, StopSel=>>') AS "preview"
FROM texts;

id=1: Wrong number of fragments (4) with highlight words are returned. 
id=2: No highlight word at all.
id=3: Correct number of fragments (6) with highlight words are returned.

... and so on. Until MaxFragments=11 where for id=1 the number of returned
fragments changes again (5)

SELECT id,
ts_headline('english', "texts"."fulltext", to_tsquery('english', 'amazon &
world'), 'MaxFragments=11, StartSel=<<, StopSel=>>') AS "preview"
FROM texts;

id=1: Wrong number of fragments (5) with highlight words are returned. 
id=2: No highlight word at all.
id=3: Correct number of fragments (11) with highlight words are returned.

... and so on. Until MaxFragments=14 where for id=2 suddenly more fragments
(2) get returned.

SELECT id,
ts_headline('english', "texts"."fulltext", to_tsquery('english', 'amazon &
world'), 'MaxFragments=14, StartSel=<<, StopSel=>>') AS "preview"
FROM texts;

id=1: Wrong number of fragments (5) with highlight words are returned. 
id=2: Wrong number of fragments (2) with highlight words are returned. 
id=3: Correct number of fragments (11) with highlight words are returned.

I stopped testing here, but im sure the strange behaviour and jumps in
fragment count will continue.

Any ideas?

[1] https://1drv.ms/u/s!AqRGi9iGBWddgZVU_M2iuoRdTzM6tg?e=0OHHnB