Conveniently bypassing the whole code/design argument, the only thing I'd do differently is to factor out the repeated code bits (i.e., I'm Lazy about typing, as well.)

sub lbl {
    my $args = { -class => 'small', -for => $_[0] };
    label($args, $_[1]);
}

sub txt {
    my $name = shift;
    my $args = { -name => $name, -id => $name };
    %$args = (%$args, %{$_[0]}) if $_[0];
    textfield($args);
}
[download]

...which would make the invocations look like this:

# Instead of
# label({-class=>'small', -for=>'ff_job_id'},'FaultFinder ID')
lbl('ff_job_id','FaultFinder ID')

# Instead of
# textfield({-name=>'searchby_status',-id=>'searchby_status'})
txt('searchby_status')

# Instead of
#textfield({-name=>'searchby_date_start',-id=>'searchby_date_start',-o
+nchange=>"checkDate(this)"})
txt('searchby_date_start', {-onchange => 'checkDate(this)'})
[download]

It's just a nitpick, but I have A Thing about retyping code.

Update: changed the usage/syntax of "txt" to be more like the usage of CGI (i.e., used a hashref instead of an arrayref); keeping the interface the same is a better methodology and leads to less confusion.

-- 
Human history becomes more and more a race between education and catastrophe. -- HG Wells