UndefinedConversionError when LANG and default encoding are different

Description

In this situation, IRB raises UndefinedConversionError

[tomoya.ishida:3.2]% echo "IRB.conf[:USE_MULTILINE]=false" > irbrc
[tomoya.ishida:3.2]% IRBRC=irbrc LANG=en_US.ASCII ruby -Eutf-8:utf-8 -e "binding.irb"              
irb(main):001> Readline::VERSION
=> "8.2"
irb(main):002> \343\201\202\343\201\202\343\201\202
/Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb/history.rb:82:in `write': "\xE3" to UTF-8 in conversion from ASCII-8BIT to UTF-8 to US-ASCII (Encoding::UndefinedConversionError)
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb/history.rb:82:in `puts'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb/history.rb:82:in `block in save_history'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb/history.rb:73:in `open'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb/history.rb:73:in `save_history'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1033:in `ensure in run'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1034:in `run'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1587:in `irb'
	from <internal:prelude>:5:in `irb'
	from -e:1:in `<main>'
/Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb/context.rb:628:in `strip': invalid byte sequence in US-ASCII (Encoding::CompatibilityError)
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb/context.rb:628:in `parse_command'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1134:in `build_statement'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1122:in `block in each_top_level_statement'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1119:in `loop'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1119:in `each_top_level_statement'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1042:in `eval_input'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1023:in `block in run'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1022:in `catch'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1022:in `run'
	from /Users/tomoya.ishida/.rbenv/versions/3.2.2/lib/ruby/gems/3.2.0/gems/irb-1.14.1/lib/irb.rb:1587:in `irb'
	from <internal:prelude>:5:in `irb'
	from -e:1:in `<main>'

Similar thing might happen in a situation described in https://github.com/ruby/reline/issues/755#issuecomment-2396392234

Result of irb_info

Ruby version: 3.2.2
IRB version: irb 1.14.1 (2024-09-25)
InputMethod: ReadlineInputMethod with ext/readline 8.2
Completion: RegexpCompletor
.irbrc paths: irbrc, /Users/tomoya.ishida/github/ruby/3.2/irbrc
RUBY_PLATFORM: arm64-darwin22
LANG env: en_US.ASCII
East Asian Ambiguous Width: 1

Terminal Emulator

Terminal.app

Setting Files

irbrc

IRB.conf[:USE_MULTILINE]=false

Reproduce the reported command with LANG=en_US.ASCII and -Eutf-8:utf-8, then inspect irb/history.rb around save_history and irb/context.rb around parse_command. Trace how the differing encodings reach history writing and command parsing. Done means the reported interaction no longer raises either encoding error.