Hi Chuck,
In your first question you mention "I see many engineers using MSL, but I believe, should be using NAVD88 as a vertical datum." I think this blog from this place might be of interest as there's a bunch of ambiguity around MSL etc...
https://community.esri.com/groups/coordinate-reference-systems/blog/2014/08/14/mean-sealevel
Regarding the next question: "what [is] the cut-off distance is between use of a tidal gauge and/or NAVD88."
With regards to the validity of a tidal gauge reading against distance from that gauge. I would treat the gauge as a single point of reference much like a survey triangulation point but arbitrary and local so maybe not like a survey triangulation point. This is because local variations in sea level are really subject to large hydrodynamic influences especially close to the shore. So, there isn't a rule of thumb as every tidal gauge would have to be seen in the context of it's local situation (e.g. topography, bathymetry, currents, etc...) perhaps a number could be achieved after hydrographic modelling. A rule of thumb would work however, it would be impossible to have a regular assumed error of vertical error based on distance from the gauge.
NAVD88 is based upon levelling of height observations across the USA and I understand that a model was built after hurricane Katrina to interpolate height, from local sample locations (i.e. tide gauges) so that an accurate model of MSL along coastal regions could be achieved. There is documentation on this on NOAA's National Geodetic Survey (NGS) but the website is unavailable until the Federal Closure ceases.
I hope this helps.