fix?: recursion error on Sanic subclass initialisation

[artcg]

Fixes #2073

Needs someone to take a look at it, not super familiar with __new__,

but it seems to fix the error and work as expected

[codecov]

Codecov Report

Merging #2072 (c9c888d) into master (15a8b5c) will decrease coverage by 0.016%.
The diff coverage is 100.000%.

@@              Coverage Diff              @@
##            master     #2072       +/-   ##
=============================================
- Coverage   92.200%   92.184%   -0.016%     
=============================================
  Files           38        38               
  Lines         3487      3480        -7     
  Branches       584       582        -2     
=============================================
- Hits          3215      3208        -7     
  Misses         184       184               
  Partials        88        88               

Impacted Files	Coverage Δ
sanic/base.py	100.000% <100.000%> (ø)
sanic/blueprints.py	100.000% <100.000%> (ø)

---

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 15a8b5c...c9c888d. Read the comment docs.

[artcg]

Would also add a test case for this if subclassing Sanic is something that is officially supported...

[ahopkins]

@artcg Thanks. Can you add a test? You can drop it at the end of test_app.py. It is something that I know is done, so we should have that in the test suite.

[ahopkins]

This works and makes sense. Nice catch.

But, I want to play with this a little more because I think it might not yet cover all the edge cases.

Let's add the test, and I will swing back on this tomorrow with some more thoughts.

[ahopkins]

Since we no longer need the comprehension, no need to have bases as a variable.

for base in BaseSanic.__bases__:

[artcg]

Is it an optimisation not just to define BaseSanic.__init__ directly?

[ahopkins]

I am pushing a change directly to this branch to remove the metaclass.