Support performance trace (#32973)

1. Add a OpenTelemetry-like shim-layer to collect traces
2. Add a simple builtin trace collector and exporter, end users could
download the diagnosis report to get the traces.

This PR's design is quite lightweight, no hard-dependency, and it is
easy to improve or remove. We can try it on gitea.com first to see
whether it works well, and fine tune the details.

---------

Co-authored-by: silverwind <me@silverwind.io>

modules/tailmsg/talimsg.go

@@ -0,0 +1,73 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package tailmsg
+
+import (
+	"sync"
+	"time"
+)
+
+type MsgRecord struct {
+	Time    time.Time
+	Content string
+}
+
+type MsgRecorder interface {
+	Record(content string)
+	GetRecords() []*MsgRecord
+}
+
+type memoryMsgRecorder struct {
+	mu    sync.RWMutex
+	msgs  []*MsgRecord
+	limit int
+}
+
+// TODO: use redis for a clustered environment
+
+func (m *memoryMsgRecorder) Record(content string) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	m.msgs = append(m.msgs, &MsgRecord{
+		Time:    time.Now(),
+		Content: content,
+	})
+	if len(m.msgs) > m.limit {
+		m.msgs = m.msgs[len(m.msgs)-m.limit:]
+	}
+}
+
+func (m *memoryMsgRecorder) GetRecords() []*MsgRecord {
+	m.mu.RLock()
+	defer m.mu.RUnlock()
+	ret := make([]*MsgRecord, len(m.msgs))
+	copy(ret, m.msgs)
+	return ret
+}
+
+func NewMsgRecorder(limit int) MsgRecorder {
+	return &memoryMsgRecorder{
+		limit: limit,
+	}
+}
+
+type Manager struct {
+	traceRecorder MsgRecorder
+	logRecorder   MsgRecorder
+}
+
+func (m *Manager) GetTraceRecorder() MsgRecorder {
+	return m.traceRecorder
+}
+
+func (m *Manager) GetLogRecorder() MsgRecorder {
+	return m.logRecorder
+}
+
+var GetManager = sync.OnceValue(func() *Manager {
+	return &Manager{
+		traceRecorder: NewMsgRecorder(100),
+		logRecorder:   NewMsgRecorder(1000),
+	}
+})