Name request log actions so they say what they mean (closes #229)

The request log's `action` values now say what they mean, as sneak named them: `request_header_too_large`, `client_is_rate_limited`, `blocked_by_block_rule` and `client_denied`, with no aliases. `banned` splits into `temporarily_banned` and `permanently_banned`, by whether the ban runs out. When a permanent ban also covers a client, that ban is the one reported.

Trips a reader: a request over a rate limit stays `client_is_rate_limited` even when the ban it makes is permanent.
Disclosure: the offence key `rule_blocked` is renamed in `clients.json` too, so an existing file holding it stops the start and must be removed. Its `version` stays 1.

Model: opus-5-5

internal/discriminator/bans.go

@@ -50,13 +50,23 @@ func (activeBan) check(_ context.Context, rq *request, now time.Time) *verdict {
 	}
 
 	return &verdict{
-		action: requestlog.ActionBanned,
+		action: banAction(ban.Permanent()),
 		figures: map[string]any{
 			figureNetblock: ban.Netblock.String(), "ban_expires": rq.line.BanExpires,
 		},
 	}
 }
 
+// banAction is the action of a request refused under a ban or for making one:
+// permanent says the ban never runs out.
+func banAction(permanent bool) string {
+	if permanent {
+		return requestlog.ActionPermanentlyBanned
+	}
+
+	return requestlog.ActionTemporarilyBanned
+}
+
 // limitFigures are the figures of a heuristic that broke hit, a limit: its
 // window, the count and the limit.
 func limitFigures(hit ratelimit.Hit) map[string]any {
@@ -151,11 +161,11 @@ func (rq *request) banForLimit(now time.Time, hit ratelimit.Hit, status int) {
 // attack, which notes name: the ban rule that matched, or the trap path
 // asked for. It fills in the rest of the notes. In observe mode it makes
 // no ban, and raises the alert for the ban it would have made, if that
-// alert would be sent.
-func (rq *request) banForAttack(now time.Time, notes bans.Notes) {
+// alert would be sent. It returns the action for that ban.
+func (rq *request) banForAttack(now time.Time, notes bans.Notes) string {
 	netblock := rq.h.netblock(rq.client)
 	if rq.h.config.Observe && !rq.wouldAlertBan(netblock, now, bans.CauseAttack) {
-		return
+		return banAction(rq.h.ledger.WouldBePermanent(netblock, now, bans.CauseAttack))
 	}
 
 	notes.ASN = rq.line.ASN
@@ -171,7 +181,7 @@ func (rq *request) banForAttack(now time.Time, notes bans.Notes) {
 			rq.alertBan(ban)
 		}
 
-		return
+		return banAction(ban.Permanent())
 	}
 
 	ban, made := rq.h.ledger.BanForAttack(netblock, now, notes)
@@ -180,16 +190,18 @@ func (rq *request) banForAttack(now time.Time, notes bans.Notes) {
 	if made {
 		rq.alertBan(ban)
 	}
+
+	return banAction(ban.Permanent())
 }
 
 // banForCrowdSec bans the client's netblock at now until decision,
 // CrowdSec's decision on the client, ends. In observe mode it makes no
 // ban, and raises the alert for the ban it would have made, if that alert
-// would be sent.
-func (rq *request) banForCrowdSec(now time.Time, decision reputation.Decision) {
+// would be sent. It returns the action for that ban.
+func (rq *request) banForCrowdSec(now time.Time, decision reputation.Decision) string {
 	netblock := rq.h.netblock(rq.client)
 	if rq.h.config.Observe && !rq.wouldAlertBan(netblock, now, bans.CauseCrowdSec) {
-		return
+		return banAction(rq.h.ledger.WouldBePermanent(netblock, now, bans.CauseCrowdSec))
 	}
 
 	notes := bans.Notes{
@@ -208,7 +220,7 @@ func (rq *request) banForCrowdSec(now time.Time, decision reputation.Decision) {
 			rq.alertBan(ban)
 		}
 
-		return
+		return banAction(ban.Permanent())
 	}
 
 	ban, made := rq.h.ledger.BanForCrowdSec(netblock, now, decision.Expires,
@@ -218,6 +230,8 @@ func (rq *request) banForCrowdSec(now time.Time, decision reputation.Decision) {
 	if made {
 		rq.alertBan(ban)
 	}
+
+	return banAction(ban.Permanent())
 }
 
 // wouldAlertBan reports whether the alert for a ban on netblock for cause