Possible suboptimal formatting on `if let` with multiple patterns

The indentation of the second pattern in these functions (especially the first one) looks kind of odd. I can look into fixing it if it should be fixed.

pub fn f(input_value: Test) -> (u8, u8) {
    let intermediate = if let Test::OptionWithLongNameA { value1, value2, .. }
    | Test::OptionWithLongNameB { value1, value2, .. } = input_value
    {
        (*value1, *value2)
    } else {
        unreachable!()
    };
    intermediate
}

pub fn g(input_value: Test) -> (u8, u8) {
    let very_very_long_name_for_intermediate =
        if let Test::OptionWithLongNameA { value1, value2, .. }
        | Test::OptionWithLongNameB { value1, value2, .. } = input_value
        {
            (*value1, *value2)
        } else {
            unreachable!()
        };
    very_very_long_name_for_intermediate
}

pub fn h(long_input_value_name: Test) -> (u8, u8) {
    if let Test::OptionWithLongNameA { value1, value2, .. }
    | Test::OptionWithLongNameB { value1, value2, .. } = long_input_value_name
    {
        (*value1, *value2)
    } else {
        unreachable!()
    }
}

enum Test {
    OptionWithLongNameA { value1: u8, value2: u8, value3: u8 },
    OptionWithLongNameB { value1: u8, value2: u8, value3: u8 },
    NeverUsed1,
}

Start by running rustfmt on the Rust examples in this issue and compare the indentation of the second pattern across functions f, g, and h. Trace the formatter behavior responsible for these pattern layouts; done means the multiple-pattern if let formatting is consistently improved without regressing the examples.