Wrong precedence in Field::getFieldSetting[s]() when setting appears in both field and instance

Indeed, Field::getSetting() looks wrong.
It should first look if the setting is present in the actual settings of the field, then fetch a 'default field-level value', then fetch a 'default instance-level value'.

Actually, getFieldSettings() will have the same logic issue.
Interdiff is against #1, contains the whole fix.

Not sure about <code>tags in simpletest messages :-)
The grammar seems a bit off too:
"Article image field instance default equals expected file ID of @fid, returned @returned by FieldInstance::getFieldSetting('default_image')"
-> "Article setting returned by FieldInstance::getFieldSetting(): @returned; expected: @fid" ?

+++ b/core/modules/image/lib/Drupal/image/Tests/ImageFieldDefaultImagesTest.phpundefined
@@ -52,7 +52,40 @@ function testDefaultImages() {
+      format_string('Article setting returned by FieldInstance::getFieldSetting(): @returned; expected: @fid.', $args)

I'm not sure how useful these assertion messages are. the expected/returned values are printed by the default output too.

The advantage of leaving it out would be that you could write the assertEqual() arguments on a single line and avoid any arguments about multiline function arguments... ;)

+++ b/core/modules/image/lib/Drupal/image/Tests/ImageFieldDefaultImagesTest.phpundefined
@@ -52,7 +52,40 @@ function testDefaultImages() {
+    // Test also FieldInstance::getFieldSetting().
...
+    // Test also Field::getFieldSettings().

Nitpick: I think this reads more fluent if you write Also test FieldInstance::getFieldSettings(). Not sure if this should use the full namespace.

Lovely. I think this is good, test is much more readable like this IMHO.

Committed/pushed to 8.x, thanks!

@claudiu.cristea: OK, I opened #2074217: Reconsider support for field-level & instance-level settings with the same name ? for discussion - and copied over the reasons why I think that is a non-issue ;-)