fix(builtins): handle quoted assignment arguments in declare

[lu-zero]

When declare receives a single-quoted argument like 'arr=()', the parser cannot recognize it as an AssignmentWord (leading quote is not a valid variable name start). After word expansion strips the quotes, the builtin receives CommandArg::String("arr=(${X})") but never detected the = sign, treating the entire string as a variable name and failing validation.

Extend declaration_to_name_and_value to split on = for CommandArg::String args, handling both name=value and name[index]=value patterns. For scalar values the expanded string is used as-is (word expansion already handled quoting correctly). For array values with -a, add lift_string_array_assignment that reconstructs a declare command and runs it through the shell to let the parser and expander handle array parsing and parameter expansion at runtime.

This fixes the Gentoo rpm.eclass pattern:
IFS=, declare -a 'types=()'

[github-actions]

Performance Benchmark Report

Benchmark name	Baseline (μs)	Test/PR (μs)	Delta (μs)	Delta %
clone_shell_object	17.25 μs	17.47 μs	0.22 μs	⚪ Unchanged
eval_arithmetic	0.15 μs	0.16 μs	0.01 μs	⚪ Unchanged
expand_one_string	1.56 μs	1.51 μs	-0.05 μs	⚪ Unchanged
for_loop	30.72 μs	29.66 μs	-1.06 μs	🟢 -3.46%
full_peg_complex	57.76 μs	57.59 μs	-0.17 μs	⚪ Unchanged
full_peg_for_loop	6.31 μs	6.30 μs	-0.01 μs	⚪ Unchanged
full_peg_nested_expansions	16.47 μs	16.34 μs	-0.13 μs	⚪ Unchanged
full_peg_pipeline	4.20 μs	4.21 μs	0.01 μs	⚪ Unchanged
full_peg_simple	1.79 μs	1.76 μs	-0.03 μs	⚪ Unchanged
function_call	3.33 μs	3.25 μs	-0.08 μs	⚪ Unchanged
instantiate_shell	52.80 μs	53.81 μs	1.01 μs	🟠 +1.91%
instantiate_shell_with_init_scripts	28227.40 μs	26160.81 μs	-2066.59 μs	🟢 -7.32%
parse_peg_bash_completion	2104.71 μs	2095.56 μs	-9.15 μs	⚪ Unchanged
parse_peg_complex	20.56 μs	19.38 μs	-1.18 μs	🟢 -5.74%
parse_peg_for_loop	2.02 μs	2.03 μs	0.00 μs	⚪ Unchanged
parse_peg_pipeline	2.06 μs	2.06 μs	-0.00 μs	⚪ Unchanged
parse_peg_simple	1.12 μs	1.10 μs	-0.01 μs	⚪ Unchanged
run_echo_builtin_command	17.19 μs	17.11 μs	-0.08 μs	⚪ Unchanged
tokenize_sample_script	3.39 μs	3.40 μs	0.00 μs	⚪ Unchanged

Code Coverage Report: Only Changed Files listed

Package	Base Coverage	New Coverage	Difference
brush-builtins/src/declare.rs	🟢 87.4%	🟢 87.47%	🟢 0.07%
brush-core/src/results.rs	🟢 83.33%	🟢 80.16%	🔴 -3.17%
Overall Coverage	🟢 75.4%	🟢 75.42%	🟢 0.02%

Minimum allowed coverage is 70%, this run produced 75.42%

Test Summary: bash-completion test suite

Outcome	Count	Percentage
✅ Pass	1594	75.58
❗️ Error	18	0.85
❌ Fail	143	6.78
⏩ Skip	339	16.07
❎ Expected Fail	13	0.62
✔️ Unexpected Pass	2	0.09
📊 Total	2109	100.00

[reubeno]

Thanks for putting this together; I hit this while looking at bash completion tests.

[Elsie19]

This can be collapsed:

Suggested change

	if let brush_core::CommandArg::String(s) = declaration {
	if let Some((_name, value)) = s.split_once('=') {
	return value.starts_with('(');
	}
	}
	if let brush_core::CommandArg::String(s) = declaration
	&& let Some((_, value)) = s.split_once('=')
	{
	return value.starts_with('(');
	}

[lu-zero]

sure, let me fold it in. Thank you :)