[ITensors] [BUG] logdot and lognorm are not compatible with AD due to try/catch

Description of bug

Due to try/catch in lognorm and logdot, they are not supported by AD.

Minimal code demonstrating the bug or unexpected behavior

l1 = x -> logdot(x, b)
l2 = x -> lognorm(x, b)
l3 = x -> norm(x, b)
inds = [Index(2) for _ in 1:10]
a = randomMPS(inds)
b = randomMPS(inds)
l1'(a)
l2'(a)
l3'(a)

Expected output or behavior

Return a valid gradient.

Actual output or behavior

Compiling Tuple{ITensors.var"##_log_or_not_dot#663", Bool, typeof(ITensors._log_or_not_dot), MPS, MPS, Bool}: try/catch is not supported.

Version information

• Output from versioninfo():

julia> versioninfo()
Julia Version 1.8.2
Commit 36034abf260 (2022-09-29 15:21 UTC)
Platform Info:
  OS: macOS (arm64-apple-darwin21.3.0)
  CPU: 10 × Apple M1 Max
  WORD_SIZE: 64
  LIBM: libopenlibm
  LLVM: libLLVM-13.0.1 (ORCJIT, apple-m1)
  Threads: 1 on 8 virtual cores

• Output from using Pkg; Pkg.status("ITensors"):

julia> using Pkg; Pkg.status("ITensors")
Status `~/.julia/environments/v1.8/Project.toml`
  [9136182c] ITensors v0.3.20

Start by reproducing the Julia snippet with ITensors v0.3.20, then inspect the logdot and lognorm implementation around _log_or_not_dot, named in the compiler error. Confirm that automatic differentiation returns valid gradients and add a regression test covering both functions.