various ed25519/ref10 api simplifications for our specific use case

ED25519 authentication plugin

* define MYSQL_DYNAMIC_PLUGIN only for server plugins
* don't typedef my_bool in mysql.h if plugin.h has already done it
* fix the include guard in plugin.h

so that auth plugins could use various thd services

move most of the code into my_sha.ic, making it independent
from the actual SHAx variant.

just as sql_plugin.cc does

because FLUSH STATUS does not reset them, so their values
are affected by previously run tests since the last server restart.

MDEV-11942 BLACKHOLE is no longer active in 10.1 by default, mysql_upgrade not handling the situation

fix the patch. add tests

Try harder to show the table's engine.

If the table's engine is not loaded, the table won't open.
But we can still read the engine name from frm as a string.

Make SELECT <columns> FROM I_S.TABLES behave identically independently
from whether <columns> require opening the table in engine or
<columns> can be filled with only opening the frm.

In particular, fill_schema_table_from_frm() should not silently skip
frms with unknown engine, but should fill the I_S.TABLES row
with NULLs just like fill_schema_table_by_open() does.

this simplifies the code and avoids unnecessary conversions back and forth.
and it works even if the engine is not installed.

this fixes the crash of innodb.innodb-blob --ps-protocol

has_no_default_value() should only fail the insert in the strict mode.

Additionally, don't check for "all fields are given values" twice,
it'll produce duplicate warnings.

If openat is present on the system and it tries to open a symlink with
O_NOFOLLOW, we get errno 40.

If openat is not present on the system, we use the alternative open call, with
slightly different logic. IF the symlink doesn't point to a valid file,
we get errno 20. This test uses an invalid symlink on the table t1.MYD.

fil_parse_write_crypt_data(): Correct the comparison operator.
This was broken in commit 498f4a82
which removed a signed/unsigned mismatch in these comparisons.

Also, implement MDEV-11027 a little differently from 5.5 and 10.0:

recv_apply_hashed_log_recs(): Change the return type back to void
(DB_SUCCESS was always returned).

Report progress also via systemd using sd_notifyf().

mysql_client uses some inline assembly code to switch thread stacks.
This works, however tools that perform backtrace get confused to fix
this we write a specific constant to signify bottom of stack. This
constant is needed when compiling with CLang as well.