Seclude general error handler and improve process execution span setup

[medikoo]

Addresses 0.1 and 0.2 from #8364

Additionally fixed tests in two unrelated files as they were immune to side effects of already required utils (ensured there runServerless runs always run with clear require cache)

[codecov]

Codecov Report

Merging #8726 (c278efe) into master (30d89d1) will increase coverage by 0.01%.
The diff coverage is 98.61%.

@@            Coverage Diff             @@
##           master    #8726      +/-   ##
==========================================
+ Coverage   87.42%   87.43%   +0.01%     
==========================================
  Files         254      256       +2     
  Lines        9615     9631      +16     
==========================================
+ Hits         8406     8421      +15     
- Misses       1209     1210       +1     

Impacted Files	Coverage Δ
lib/classes/Error.js	96.00% <ø> (-2.64%)	⬇️
lib/cli/handle-error.js	98.38% <98.38%> (ø)
lib/Serverless.js	97.19% <100.00%> (-0.13%)	⬇️
lib/cli/resolve-local-serverless-path.js	100.00% <100.00%> (ø)
lib/classes/CLI.js	93.03% <0.00%> (+0.03%)	⬆️

---

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 30d89d1...c278efe. Read the comment docs.

[pgrzesik]

One comment about duplicated method, otherwise looks great 👍

[pgrzesik]

This is duplicated across two modules or I'm missing the difference between them?

[medikoo]

Good point, I didn't notice (linter simply told me it's still used in Error.js so I didn't touch it).

As I see this logic is also used in logWarning, and it's the reason it's also kept there.

I'm not sure is it worth to clean this up, as (1) logWarning and all other log functions will be removed with #1720 which is also on priority list, and (2) handle-error.js as introduced here is also likely to be improved (in context of this PR we simply take this logic out of Serverless class context, so Serverless is not mangled with CLI specific logic)

It's likely that quite soon writeMessage will be gone from both locations (I also personally don't see it as part of cli utils), and investing time now to make a tested reusable util from it now, doesn't feel that justified.

Let me know

[pgrzesik]

If we'll be removing it soon I see zero harm in keeping in duplicated for this short period of time 👍