This PR modifies how big and small values are formatted by WKTWriter by switching from positional to scientific notation where necessary, to avoid rounding to zero or overflowing a char buffer crash.

This somewhat expands the "trim" mode definition to mean "print pretty numbers" since it doesn't always honour (e.g.) "3" to show fixed millimetre precision, since it will use scientific notation to show smaller non-zero numbers if necessary. The non-trim mode still uses fixed notation, and will show (e.g.) "-0" in some cases.

Also the char buffer used for these numbers is reduced from 128 to 28.

Some new unit tests are added for GEOS_printDouble from the CAPI. Other unit tests check outputs with trim set on/off.

Closes #970

Some [unpractical] geometries mapped in SI meters:

1. Classical electron radius: ./bin/geosop -a "point(0 0)" buffer 2.8179403227e-15

POLYGON ((2.8179403227e-15 0, 2.763794389558889e-15 -5.497528849777754e-16, 2.6034373879807804e-15 -1.0783790748908248e-15, 2.3430317476070782e-15 -1.5655637617177719e-15, 1.992584711160178e-15 -1.9925847111601778e-15, 1.5655637617177723e-15 -2.3430317476070782e-15, 1.078379074890825e-15 -2.6034373879807804e-15, 5.497528849777756e-16 -2.763794389558889e-15, 1.7254907981914074e-31 -2.8179403227e-15, -5.497528849777752e-16 -2.763794389558889e-15, -1.0783790748908246e-15 -2.6034373879807804e-15, -1.5655637617177713e-15 -2.343031747607079e-15, -1.9925847111601778e-15 -1.992584711160178e-15, -2.3430317476070786e-15 -1.5655637617177719e-15, -2.6034373879807804e-15 -1.078379074890825e-15, -2.763794389558889e-15 -5.497528849777764e-16, -2.8179403227e-15 -3.4509815963828148e-31, -2.763794389558889e-15 5.497528849777757e-16, -2.6034373879807808e-15 1.0783790748908244e-15, -2.343031747607079e-15 1.5655637617177713e-15, -1.9925847111601786e-15 1.9925847111601778e-15, -1.5655637617177719e-15 2.3430317476070782e-15, -1.0783790748908263e-15 2.60343738798078e-15, -5.497528849777765e-16 2.763794389558889e-15, -5.176472394574221e-31 2.8179403227e-15, 5.497528849777755e-16 2.763794389558889e-15, 1.0783790748908253e-15 2.6034373879807804e-15, 1.5655637617177711e-15 2.343031747607079e-15, 1.9925847111601774e-15 1.9925847111601786e-15, 2.3430317476070782e-15 1.5655637617177719e-15, 2.60343738798078e-15 1.0783790748908265e-15, 2.763794389558889e-15 5.497528849777767e-16, 2.8179403227e-15 0))

2. Observable universe: ./bin/geosop -a "point(0 0)" buffer 4.40e26

POLYGON ((4.4e+26 0, 4.315455233774214e+26 -8.583974168709644e+25, 4.0650699430496615e+26 -1.683807102406395e+26, 3.658466294131199e+26 -2.4445090252862496e+26, 3.111269837220809e+26 -3.111269837220809e+26, 2.4445090252862503e+26 -3.658466294131199e+26, 1.683807102406395e+26 -4.0650699430496615e+26, 8.583974168709647e+25 -4.315455233774214e+26, 26942229581.24177 -4.4e+26, -8.58397416870964e+25 -4.315455233774214e+26, -1.6838071024063948e+26 -4.0650699430496615e+26, -2.4445090252862486e+26 -3.6584662941312e+26, -3.111269837220809e+26 -3.111269837220809e+26, -3.6584662941312e+26 -2.4445090252862496e+26, -4.0650699430496615e+26 -1.6838071024063955e+26, -4.315455233774214e+26 -8.583974168709659e+25, -4.4e+26 -53884459162.48354, -4.315455233774214e+26 8.583974168709649e+25, -4.065069943049662e+26 1.6838071024063945e+26, -3.6584662941312e+26 2.4445090252862486e+26, -3.11126983722081e+26 3.111269837220809e+26, -2.4445090252862496e+26 3.658466294131199e+26, -1.6838071024063976e+26 4.065069943049661e+26, -8.58397416870966e+25 4.3154552337742135e+26, -80826688743.72531 4.4e+26, 8.583974168709645e+25 4.315455233774214e+26, 1.6838071024063962e+26 4.0650699430496615e+26, 2.4445090252862482e+26 3.6584662941312e+26, 3.1112698372208085e+26 3.11126983722081e+26, 3.658466294131199e+26 2.4445090252862496e+26, 4.065069943049661e+26 1.683807102406398e+26, 4.3154552337742135e+26 8.583974168709664e+25, 4.4e+26 0))

3. The finite extremes of double-precision: ./bin/geosop -a "point(5e-324 1.7e308)"

POINT (5e-324 1.7e+308)