fix: anthropic tool-call round-trip, cache zero-hit parity, round-robin load balancing

anthropic.lua v3.0.0:
- Issue 1: tool_result/tool_use round-trip
- Issue 3: thinking default OFF (opt-in via extra_body.thinking)
- Issue 4: tool_choice mapping
- Issue 5: collect_blocks preserves unknown part types
- message_stop no longer emits done=true (was overwriting tool_calls finish_reason)
- cache_read_input_tokens normalized even at 0

gemini.lua:
- transform_response was missing cachedContentTokenCount

openai.lua (Issue 6):
- transform_error handles flat envelopes, nginx HTML, bare text

chat.go mergeUsage:
- Keep PromptTokensDetails even when CachedTokens=0

scheduler.go:
- Remove sort.SliceStable by Pref; round-robin cursor is the only LB mechanism

provider.go ModelAvailable:
- Also check Pref() > prefMin, persistently failing slots exit cands

presets.go:
- 17 built-in source templates

Tests: 6 new test functions, 2 updated for new semantics

internal/gateway/chat.go

@@ -671,7 +671,11 @@ func mergeUsage(prev, cur *types.TokenUsage) *types.TokenUsage {
 		out.Completion = cur.Completion
 	}
 	out.Total = out.Prompt + out.Completion
-	if cur.PromptTokensDetails != nil && cur.PromptTokensDetails.CachedTokens > 0 {
+	if cur.PromptTokensDetails != nil {
+		// Keep the details object even when CachedTokens is 0: a reported
+		// zero-hit is meaningful ("cache missed") and must stay
+		// distinguishable from "upstream never reported cache info".
+		// Dropping it here made streaming rows show “—” instead of 0%.
 		out.PromptTokensDetails = cur.PromptTokensDetails
 	}
 	if cur.PromptCacheHit > 0 {