Implement the new entity field API for the number field type

Here's a first shot.

Used string as the value for DecimalItem. I guess we could add a specific TypedData class, not sure.

See #1777268-3: Add a TypedData API for allowing metadata to be associated with data and following about the whole Decimal / Float debate. I don't really know much about the whole thing, but I remembered that discussion so wanted to give a pointer, because I'm pretty sure it's related at the very least.

Thanks. So that discussion essentially said that we do not want to add a Decimal type. Which means use string here as we did not have before and still don't have a native interpretation of this in PHP...

On the other side Language isn't a native data type in PHP either :)

Updates for Contrib module one	Other	LocaleUpdateTest.php	436	Drupal\locale\Tests\LocaleUpdateTest->testUpdateImportSourceRemote()

Haven't seen that one before..

#1: number-field-item-1839076-1.patch queued for re-testing.

Updated to use static::

Patch looks gread and has solid test coverage - good work!
-> RTBC as long as it still applies after the comments patch was committed.

#7: number-field-item-1839076-7.patch queued for re-testing.

No idea why this failed to apply.

Anyway, here is a re-roll against #1839082: Implement the new entity field API for the field test field type so that I can use the new base test class. The patch for the testbot contains the changes from that other issue so he can run the tests.

+++ b/core/modules/number/lib/Drupal/number/Tests/NumberItemTest.php
@@ -0,0 +1,106 @@
+    $float = rand(0, 1000) / 100;
...
+    $new_float = rand(1001, 2000) / 100;

I know this is very nitpicky, but shouldn't we ensure there's always at least one decimal digit? Otherwise this could be a candidate for a mandelbug, e.g. if somewhere the handling of decimal digits fails or vice versa.
Could be done like this:

  do {
    $float = rand(0, 1000) / 100;
  } while(fmod($float, 1) == 0);

#12: number-field-item-1839076-7.patch queued for re-testing.

@das-peter: A float without decimal places seems like a valid value. I'll make sure that it doesn't cause any issues with an explicit test run and confirm but I don't think we need to treat that specially.

Re-roll that uses a fixed float value that does not cause any trouble. We have to fight with enough random fails already :)

Thanks. Patch looks good and das-peter's concerns have been addressed.

Committed to 8.x. Thanks.