A difficult question (and quite refreshing, I'm on a gap year and trying to get back into this for upcoming interviews). Finding z(t) involves solving the second order differential equation you end up with in part one. The general solution for an equation of this type (as you probably know) is:
z(t)=Aeλ1 t+Beλ2 t+C
I'll call this (1)
Where λ1 and λ2 are the solutions of the auxilliary equation:
λ²-ω²sin²θ=0
(so λ1=ωsinθ and λ2=-ωsinθ )
The constants take a bit of work to determine but it's all fairly straightforward algebra: Differentiate (1) to get z.(t), and use z.(0)=0 to arrive at A=B. Then use z(0)=h in the (1) to get 2A+C=h. Thirdly, use the equation arrived at in the first part, put z(0)=h into it and conclude that z: (0)=hω²sin²θ-gcos²θ. Differentiate z.(t) to get another form z: (t) and put z: (0)=hω²sin²θ-gcos²θ to hopefully get your three constants A, B and C. Let me know if you need any of this expanding on.
You should arrive at A=B=½h-(gcos²θ/2ω²sin²θ ) or ½h-(gcot²θ/2ω²) and C=gcos²θ/ω²sin²θ
So your final z(t) is defined as
z=(½h-(gcot²θ/2ω²))(eωtsinθ+e-ωtsinθ)+gcos²θ/ω²sin²θ
I'll call this (2)
I've checked that and I think it all works out, but I am known to slip up occasionally. I'll know for sure when I've done the rest of it, bear with me.
Edit: In fact, I think I have made a mistake. z.(t) needs to be a function involving h also and according to the above, it isn't. I'll try and find my error.
Edit2: I put z(0)=0 instead of z(0)=h, it's right now I think.
Edit3: Made another stupid error (left out a squared sign), I've corrected this post and now the very last part is fairly trivial:
Looking at (2) in pieces:
(eωtsinθ+e-ωtsinθ)
This part gets more positive with increasing time, since the first half just increases exponentially and the other half gets increasingly tiny (and remains positive).
+gcos²θ/ω²sin²θ
This is a constant, negative because g is negative and the rest is squared. Although it is negative it becomes less significant as t increases since the magnitude of the first half of (2) increases rapidly.
(½h-(gcot²θ/2ω²)) or ½(h-(gcot²θ/ω²))
This is the crux of the matter, the sign of z as t becomes large rests here because it is constant and it multiplies the significant (and positive) part of (2). Obviously, if h>gcot²θ/ω² then this constant is positive and the whole thing will get increasingly positive (z tends to ∞ ), similarly if h<gcot²θ/ω² then z tends to -∞.