Copying parser/lexer

[David Kastrup]

Hi,

when working on the embedded Scheme stuff I got a nasty surprise working
with cloned lexers/parsers.

They start out with the error flag of their source.

That's stupid, since it means that they can't complete successfully
without explicitly clearing their respective error flags.

I got hit by this since tests complained about every single #{ ... #}
expression when a version error in the main file set the error flag
right at the start of the run.

As we don't clone the parsers for fun (or continuation), I suggest
starting the clones with reset error state right in the copy
constructor.

If I don't encounter sufficient resistance, I'll likely push a change
like that to dev/staging tomorrow.

-- 
David Kastrup