Guard the UnixTime globals, so JsonWebToken and Security can be used together

[mckuipers]

JWT/UnixTime.pkg guards five of its definitions — gVoid, tFileTime, tSystemTime, and
both External_Functions — but not its two global functions. This adds the missing guards.

Two commits, separable: the first fixes a real collision, the second is consistency.

Commit 1 — CurrentUnixTime, which actually clashes

DataFlex-dev/Security ships a byte-identical copy of this file as include/UnixTime.pkg,
also unguarded. DataFlex resolves a Use by path, so to the compiler these are two unrelated
files, and any program containing both gets the global twice:

Error 4390: Illegal method name definition
            Global method GET CURRENTUNIXTIME already defined

That combination is not exotic — it is what the OAuth Server Lib needs. OAuth Server Lib
requires this package, and any OIDC provider offering TOTP as a second factor also wants
cSecureOneTimePassword from Security. The first build that puts both in one program fails,
and the error points into a package-manager-installed folder that cannot be edited, so it reads
like a broken dependency rather than a name collision.

Reproduce by building any program that uses cSecureOneTimePassword and JWT.pkg together.

The matching PR is open on DataFlex-dev/Security. Guarding either copy fixes the collision;
both are being offered one so neither library depends on Use order.

Commit 2 — UnixTimeToDateTime, for consistency

Not needed for the collision above: Security's copy does not define UnixTimeToDateTime, so
that name does not clash between the two libraries today. It makes the file idempotent —
including it a second time by a different path becomes a no-op rather than an error — which is
what the six guards around it already achieve for everything else.

Drop this commit if you would rather change only the name that actually clashes.

DateTimetoUnixTime deliberately gets no guard

It is declared without the Global keyword, and the test below shows it does not collide.

Why Get_<Name> is the right symbol

A Function <Name> Global defines Get_<Name>. The precedent is DataFlex's own
cWorkspace.pkg, which guards Function LastDelimeter Global with #IFDEF Get_LastDelimeter.

I used this file's existing #IFNDEF … #ENDIF style rather than cWorkspace.pkg's
#IFDEF … #ELSE … #ENDIF, to match the guards already here.

Verification

Compiled this file into a program that already contains the whole package, so every global in
it is defined twice by two different paths:

	Result
unpatched	4390 at (40,1) GET CURRENTUNIXTIME
commit 1 only	4390 at (59,1) GET UNIXTIMETODATETIME
both commits	clean compile

DateTimetoUnixTime was double-included in all three runs and never raised 4390, which is
the basis for leaving it alone. The compiler reports all 4390s in a run rather than stopping at
the first, so its absence is evidence and not an artefact of ordering.

The same build asserts the symbol name is right rather than assuming it:

#IFNDEF Get_CurrentUnixTime
    #ERROR 999 Get_CurrentUnixTime is NOT the symbol
#ENDIF
#IFDEF Get_CurrentUnixTimeNeverDefinedAnywhere
    #ERROR 998 IFDEF matched an undefined symbol
#ENDIF

placed after a known definition. Neither #ERROR fires — the second exists so the first cannot
pass vacuously.