Add C++20 char8_t/u8string support (#2026)

* Fix test build in C++20

* Add C++20 char8_t/u8string support
diff --git a/tests/test_builtin_casters.py b/tests/test_builtin_casters.py
index abbfcec..9142258 100644
--- a/tests/test_builtin_casters.py
+++ b/tests/test_builtin_casters.py
@@ -15,6 +15,8 @@
     assert m.good_utf16_string() == u"bβ€½πŸŽ‚π€z"
     assert m.good_utf32_string() == u"aπ€πŸŽ‚β€½z"
     assert m.good_wchar_string() == u"aβΈ˜π€z"
+    if hasattr(m, "has_u8string"):
+        assert m.good_utf8_u8string() == u"Say utf8β€½ πŸŽ‚ 𝐀"
 
     with pytest.raises(UnicodeDecodeError):
         m.bad_utf8_string()
@@ -29,12 +31,17 @@
     if hasattr(m, "bad_wchar_string"):
         with pytest.raises(UnicodeDecodeError):
             m.bad_wchar_string()
+    if hasattr(m, "has_u8string"):
+        with pytest.raises(UnicodeDecodeError):
+            m.bad_utf8_u8string()
 
     assert m.u8_Z() == 'Z'
     assert m.u8_eacute() == u'é'
     assert m.u16_ibang() == u'β€½'
     assert m.u32_mathbfA() == u'𝐀'
     assert m.wchar_heart() == u'♥'
+    if hasattr(m, "has_u8string"):
+        assert m.u8_char8_Z() == 'Z'
 
 
 def test_single_char_arguments():
@@ -92,6 +99,17 @@
         assert m.ord_wchar(u'aa')
     assert str(excinfo.value) == toolong_message
 
+    if hasattr(m, "has_u8string"):
+        assert m.ord_char8(u'a') == 0x61  # simple ASCII
+        assert m.ord_char8_lv(u'b') == 0x62
+        assert m.ord_char8(u'é') == 0xE9  # requires 2 bytes in utf-8, but can be stuffed in a char
+        with pytest.raises(ValueError) as excinfo:
+            assert m.ord_char8(u'Δ€') == 0x100  # requires 2 bytes, doesn't fit in a char
+        assert str(excinfo.value) == toobig_message(0x100)
+        with pytest.raises(ValueError) as excinfo:
+            assert m.ord_char8(u'ab')
+        assert str(excinfo.value) == toolong_message
+
 
 def test_bytes_to_string():
     """Tests the ability to pass bytes to C++ string-accepting functions.  Note that this is
@@ -116,10 +134,15 @@
     assert m.string_view_chars("Hi πŸŽ‚") == [72, 105, 32, 0xf0, 0x9f, 0x8e, 0x82]
     assert m.string_view16_chars("Hi πŸŽ‚") == [72, 105, 32, 0xd83c, 0xdf82]
     assert m.string_view32_chars("Hi πŸŽ‚") == [72, 105, 32, 127874]
+    if hasattr(m, "has_u8string"):
+        assert m.string_view8_chars("Hi") == [72, 105]
+        assert m.string_view8_chars("Hi πŸŽ‚") == [72, 105, 32, 0xf0, 0x9f, 0x8e, 0x82]
 
     assert m.string_view_return() == "utf8 secret πŸŽ‚"
     assert m.string_view16_return() == "utf16 secret πŸŽ‚"
     assert m.string_view32_return() == "utf32 secret πŸŽ‚"
+    if hasattr(m, "has_u8string"):
+        assert m.string_view8_return() == "utf8 secret πŸŽ‚"
 
     with capture:
         m.string_view_print("Hi")
@@ -132,6 +155,14 @@
         utf16 πŸŽ‚ 8
         utf32 πŸŽ‚ 7
     """
+    if hasattr(m, "has_u8string"):
+        with capture:
+            m.string_view8_print("Hi")
+            m.string_view8_print("utf8 πŸŽ‚")
+        assert capture == """
+            Hi 2
+            utf8 πŸŽ‚ 9
+        """
 
     with capture:
         m.string_view_print("Hi, ascii")
@@ -144,6 +175,14 @@
         Hi, utf16 πŸŽ‚ 12
         Hi, utf32 πŸŽ‚ 11
     """
+    if hasattr(m, "has_u8string"):
+        with capture:
+            m.string_view8_print("Hi, ascii")
+            m.string_view8_print("Hi, utf8 πŸŽ‚")
+        assert capture == """
+            Hi, ascii 9
+            Hi, utf8 πŸŽ‚ 13
+        """
 
 
 def test_integer_casting():