outposts/ldap: fix race condition when refreshing the provider

Fixes the race condition causing the crash found in #4138, which doesn't
actually have anything to do with the issue itself.

As far as I can work out, when the outpost refreshes its list of
providers, it copies over its `boundUsers`, probably to avoid having to
fetch them all again, and does so by making a shallow copy of that
`map`, but not the mutex associated with it. It now has multiple
references to the same map, each protected by a different mutex, which
under certain conditions can cause a `concurrent map read and map write`
error.

This fix copies the map contents instead of make a shallow copy.

Signed-off-by: Marc 'risson' Schmitt <marc.schmitt@risson.space>

internal/outpost/ldap/instance.go

@@ -9,6 +9,7 @@ import (
 	"github.com/go-openapi/strfmt"
 	"github.com/nmcclain/ldap"
 	log "github.com/sirupsen/logrus"
+
 	"goauthentik.io/api/v3"
 	"goauthentik.io/internal/constants"
 	"goauthentik.io/internal/outpost/ldap/bind"
@@ -39,7 +40,7 @@ type ProviderInstance struct {
 	outpostName         string
 	outpostPk           int32
 	searchAllowedGroups []*strfmt.UUID
-	boundUsersMutex     sync.RWMutex
+	boundUsersMutex     *sync.RWMutex
 	boundUsers          map[string]*flags.UserFlags
 
 	uidStartNumber int32