[alexeid]

Several of BEAUti's default priors are improper, and so is the hyperprior it adds. This PR replaces them with a single proper default, LogNormal(M = 0, S = 8) in log space (median 1, 95% from 1.6×10⁻⁷ to 6.4×10⁶), which is wide enough for any time unit. BEAST 3's BEAUti gets the same defaults (CompEvol/beast3#208).

Default priors

Prior	Was	Now
strict clock rate (Standard.xml)	Uniform(0, ∞), improper	LogNormal(0, 8)
clock rate, relaxed mean rate (ClockModels.xml)	Gamma(0.01, 100)	LogNormal(0, 8)
mutation rate (Standard.xml, SubstModels.xml)	OneOnX, improper	LogNormal(0, 8)
Yule birth rate	Uniform(0, ∞), improper	LogNormal(0, 8)
calibrated-Yule and birth-death birth rates	Uniform(0, 1000)	LogNormal(0, 8)
constant and exponential coalescent population size	OneOnX, improper	LogNormal(0, 8)
EBSP populationMean	Gamma(0.01, 100)	LogNormal(0, 8)
Bayesian skyline, first population size	Jeffreys (1/x), improper	exponential with an estimated mean, bPopSizesInitialMean: LogNormal(0, 8) prior, scale operator, trace-log column

Gamma(0.01, 100) is proper, but about half its mass lies below 10⁻²⁸. In practice it's a 1/x prior with an arbitrary cutoff far below any meaningful value. The birth–death relative death rate (Uniform(0, 1)) and the other proper defaults are unchanged.

Hyperprior

Ticking "estimate" on a prior's own parameter added a OneOnX hyperprior with a scale operator. That's improper, and a scale operator can't move a value of 0, such as a LogNormal's M = 0. Now:

• A positive parameter (lower bound ≥ 0 or value > 0) gets LogNormal(0, 8) with a scale operator.
• Any other gets Normal(0, 8) with a random-walk operator.

Legacy parameters often have no lower bound even when they're positive (a LogNormal's S, for example), hence the value test.

Testing

• I compiled the changed classes against BEAST.base 2.7.9 and BEAST.app 2.7.8. With them and these templates, beauti -nex anolis.nex -exitaction writexml writes a Yule analysis with the new birth-rate prior, and it runs.
• I didn't run the TestFX GUI tests (ant isn't installed here, and on macOS the tests drive the real mouse). So the non-default templates (skyline, coalescent, birth–death, relaxed clock) and the hyperprior dialog haven't been exercised in the GUI. Please run the BeastFX tests, or click through these before merging.
• The same model changes, run as BEAST 2 examples in Give the calibration and tree-prior examples proper priors beast2#1225, match BEAST 3 in all ten examples.