Test over OTP 27 and rebar3 3.23

[paulo-ferraz-oliveira]

Notes on pull request

Found issues are around the development and maintenance of this library and rebar3_format. They do not seem to indicate elvis_core as non- OTP-27-ready.

[paulo-ferraz-oliveira]

Since this is now forbidden in OTP 27

===> Compiling elvis_core
compile: warnings being treated as errors
compile: warnings being treated as errors
/home/runner/work/elvis_core/elvis_core/test/examples/fail_no_successive_maps.erl:6:22: expression updates a literal
%    6|     M = #{this => is}#{wrong => "and"},
%     |                      ^

/home/runner/work/elvis_core/elvis_core/test/examples/pass_no_successive_maps_elvis_attr.erl:8:22: expression updates a literal
%    8|     M = #{this => is}#{wrong => "and"},
%     |                      ^

we might have to tweak the tests (macros 🎉) and think about dropping this check in the future 😄

[paulo-ferraz-oliveira]

rebar3 format --verify fails for OTP 24, but if I format the files with OTP 24 it then fails for OTP 27 😕

[paulo-ferraz-oliveira]

To move forward, I propose (thoughts?) we move the minimum supported version to 25.

At the same time, though, the formatter moves -else. to - else . which is consumable but not pretty 😄

I propose CI only checks formatting when OTP 26, since 27 breaks it (actually, it seems 25 already broke it). This is an important change in Erlang/OTP in any case, but an issue for rebar3_format, not this repository (you already have AdRoll/rebar3_format#359).