fix(android): fix crash when using window.add([]) with a null object

[m1ga]

Issue
If you add a null object when using .add([]) it will crash with

(main) [178,178] <embedded>:5351
    _add.call(this, child);
         ^
Error: Attempt to invoke virtual method 'java.lang.Class java.lang.Object.getClass()' on a null object reference
    at Window.add (<embedded>:5351:10)
    at new Controller (/alloy/controllers/index.js:93:5)
    at Object.exports.createController (/alloy.js:427:8)
    at /app.js:32:7
    at Module._runScript (ti:/kroll.js:1196:15)
    at Module.load (ti:/kroll.js:734:13)
    at Module.loadJavascriptText (ti:/kroll.js:1062:15)
    at Module.loadAsFile (ti:/kroll.js:1108:22)
    at Module.loadAsFileOrDirectory (ti:/kroll.js:1039:26)
    at Module.require (ti:/kroll.js:865:30)

    org.appcelerator.titanium.proxy.TiViewProxy.add(TiViewProxy.java:590)

because it tries to get the class name for the warning. Since it is null it will crash instead of showing a warning.

This PR adds a null check and outputs a different warning.

Test

const win = Ti.UI.createWindow({layout:"vertical"});
const lbl1 = Ti.UI.createLabel({text:"lbl"});
const lbl2 = Ti.UI.createLabel({text:"lbl"});
win.add([lbl1, , lbl2]);
win.open();

[hansemannn]

Why would someone want to do that? It feels like the correct behavior. We shouldn't need to save-guard everything and educate devs to properly null-check their view hierarchies.

[hansemannn]

I am not in favor of additional logic in this performance critical call

[m1ga]

It won't print the warning what happened because it will crash. This way it will give the dev the warning. The check is already in the fail part so it shouldn't be called in a normal app.

But when a user is adding dynamic elements (I only used the blank in my example to quickly test it but it could also be an empty variable) using an array it might appear that one of those is null. Rather then crashing I thought it would be nice to give the proper warning.

[hansemannn]

Maybe someone else can vote for it? I would not add this, but if others want it, we can include it.

[m1ga]

it's been like this for 9 years and no complaints, so it's not a big issue at all. So I'm fine with just closing it.
I just had it before when I was preparing another demo for a PR and was curious why it it was crashing 😄 But I accidentally added ,, in my array.

[hansemannn]

Let's take this, as you've correctly mentioned that the extra if case is only processed if there is an error case already.