fix: add encoding='utf-8', errors='replace' to subprocess.Popen

[shushuzn]

Fix: Windows GBK encoding error in browser tool

Problem: On Windows, subprocess.Popen defaults to system encoding (GBK).
When reading browser command output from GitHub pages (UTF-8 content), Python
fails with:

'gbk' codec can't decode byte 0xa6 in position 1068: illegal multibyte sequence

Solution: Add encoding='utf-8', errors='replace' as kwargs to
subprocess.Popen in _run_browser_command.

Files changed: Only tools/browser_tool.py — adds two lines:

proc = subprocess.Popen(
    ...,
    encoding='utf-8',   # NEW
    errors='replace',  # NEW
)

Testing: Verified on Windows — browser_navigate + browser_console
correctly return UTF-8 Chinese content from GitHub pages.

[teknium1]

Thanks for the Windows encoding report and focused patch.

• Current main writes browser subprocess output to raw stdout/stderr files at tools/browser_tool.py:2444-2450 and performs the decode explicitly at tools/browser_tool.py:2477-2480 with encoding="utf-8".
• This removes dependence on the Windows system locale for the UTF-8 browser output described in the report.
• The current implementation was present in d682f320b35a13084371a541a835e1d988c982b8; the proposed Popen location has since been superseded by that file-output path.

Closing as implemented on main. This is an automated hermes-sweeper review.