gegen normalize does not realize that a new variable shall be created

[felixholub]

Describe the bug
When using gegen with normalize, a target exists error is returned if a variable beginning with the same name exists.
With mean, for example, this does not happen.

Code Sample

sysuse auto, clear

// works fine with mean
gegen mpg_aa = mean(mpg)
gegen mpg_a = mean(mpg)

// does not work with normalize
gegen mpg_bb = normalize(mpg)
// throws arrow as it thinks that mpg_b refers to mpg_bb
gegen mpg_b = normalize(mpg), verbose bench(3)

. sysuse auto, clear
(1978 Automobile Data)

. // works fine with mean
. gegen mpg_aa = mean(mpg)
. gegen mpg_a = mean(mpg)

. // does not work with normalize
. gegen mpg_bb = normalize(mpg)

. // throws arrow as it thinks that mpg_b refers to mpg_bb
. gegen mpg_b = normalize(mpg), verbose bench(3)
Parsed by variables; .002 seconds
gstats_transform: target mpg_b exists without replace
r(198);

Version info

• OS: Windows 10
• Version: 1.7.5

[mcaceresb]

@felixholub As a quick fix, add set varabbrev off

I legitimately cannot remember what issues varabbrev caused that effected me to comment out this line. I am personally very strongly against varabbrev but have had to deal with it somewhat since Stata has it turned on by default.

I believe the fix is to change the way I check for whether a variable exists from a failure of confirm var to a success of confirm new var, but won't have time to test it for a bit.