Update Go to 1.26 and upgrade all dependencies

[anmarchenko]

Summary

• Update Go version from 1.24 to 1.26 across go.mod, CI workflows, and AGENTS.md
• Upgrade all direct and indirect dependencies to their latest versions (notably dd-trace-go/v2 v2.4.1 → v2.6.0, datadog-agent/* v0.73.1 → v0.76.1, gopsutil/v4 v4.25.11 → v4.26.2, grpc v1.77.0 → v1.79.1)

E2E testing

Only basic verification is required: check that parallelization behaves without any changes

[anmarchenko]

E2E Test Report: SUCCESS ✅

Tested by: Shepherd Agent (autonomous QA for Datadog Test Optimization)

Test Environment

• Method: Local testing with sidekiq playground (ddtest parallel mode against mockdog)
• Revision: 5233dbb

Results

Check	PR Branch	Baseline (main)
Total events	585	585
Tests	423	423
Suites	154	154
Modules	4	4
Sessions (parallel runners)	4	4
Test files split	36	36
Error spans	30	32
Telemetry	valid	valid
dd-trace-go version	v2.6.0	v2.4.1

Verification

Check	Status
ddtest builds with Go 1.26	✅
Parallelization (4 runners)	✅
Test file splitting (36 files)	✅
CI test cycle events submitted	✅
Telemetry valid	✅
No regressions vs baseline	✅

Test Methodology

1. Built ddtest from PR branch anmarchenko/update_go_and_deps (Go 1.26, dd-trace-go v2.6.0)
2. Ran sidekiq tests in parallel mode (--min-parallelism 1 --max-parallelism 4) against local mockdog
3. Ran identical baseline test with ddtest main branch for comparison
4. Compared event counts, parallelization behavior, and error rates
5. Confirmed all failures are pre-existing (Redis contention in parallel mode), not caused by the PR

Notes

• Pre-existing test failures (30 vs 32 error spans) are caused by Redis contention when running sidekiq tests in parallel — tests share Redis db 15. These failures are non-deterministic and appear in both PR and baseline runs.

---

This E2E test was performed by Shepherd — autonomous QA agent for Datadog Test Optimization