remove sciencedirect links for DOIs are used instead (#737)

this seems to work well with the linkcheck

docs/source/refs.bib

@@ -122,7 +122,6 @@ @article{langanke:2001
 year = {2001},
 issn = {0092-640X},
 doi = {10.1006/adnd.2001.0865},
-url = {https://www.sciencedirect.com/science/article/pii/S0092640X01908654},
 author = {K. Langanke and G. Martínez-Pinedo},
 abstract = {The weak interaction rates in stellar environments are computed for pf-shell nuclei in the mass range A=45–65 using large-scale shell-model calculations. The calculated capture and decay rates take into consideration the latest experimental energy levels and log ft-values. The rates are tabulated at the same grid points of density and temperature as those used by Fuller, Fowler, and Newman for densities ρY e =10–1011 g/cm3 and temperatures T=107–1011 K, and hence are relevant for both types of supernovae (Type Ia and Type II). Effective 〈ft〉 values for capture rates and average neutrino (antineutrino) energies are also given to facilitate the use of interpolated rates in stellar evolution codes.}
 }
@@ -150,7 +149,6 @@ @article{seitenzahl:2009
 year = {2009},
 issn = {0092-640X},
 doi = {10.1016/j.adt.2008.08.001},
-url = {https://www.sciencedirect.com/science/article/pii/S0092640X0800051X},
 author = {Ivo R. Seitenzahl and Dean M. Townsley and Fang Peng and James W. Truran},
 abstract = {We solve the equations of nuclear statistical equilibrium (NSE) for the mass fractions of 443 nuclides, including the effects of temperature dependent nuclear partition functions [T. Rauscher, F. Thielemann, At. Data Nucl. Data Tables 75 (2000) 1–351] and Coulomb corrections [G. Chabrier, A.Y. Potekhin, Phys. Rev. E 58 (1998) 4941]. The resulting mass fractions are then convolved with the new weak interaction rates for pf-shell nuclei based on large-scale shell model calculations [K. Langanke, G. Martínez-Pinedo, At. Data Nucl. Data Tables 79 (2001) 1–46] to get the rate of neutronization and the specific neutrino luminosity of matter in NSE. We present tables of the results and give examples of how the tables can be used in Type Ia supernova simulations.}
 }
@@ -296,7 +294,6 @@ @article{pepiot-desjardins:2008
 year = {2008},
 issn = {0010-2180},
 doi = {10.1016/j.combustflame.2007.10.020},
-url = {https://www.sciencedirect.com/science/article/pii/S0010218007003264},
 author = {P. Pepiot-Desjardins and H. Pitsch},
 keywords = {DRGEP, Chemical kinetics reduction, Skeletal chemistry, Autoignition, Iso-octane},
 abstract = {Production rates obtained from a detailed chemical mechanism are analyzed in order to quantify the coupling between the various species and reactions involved. These interactions can be represented by a directed relation graph. A geometric error propagation strategy applied to this graph accurately identifies the dependencies of specified targets and creates a set of increasingly simplified kinetic schemes containing only the chemical paths deemed the most important for the targets. An integrity check is performed concurrently with the reduction process to avoid truncated chemical paths and mass accumulation in intermediate species. The quality of a given skeletal model is assessed through the magnitude of the errors introduced in the target predictions. The applied error evaluation is variable-dependent and unambiguous for unsteady problems. The technique yields overall monotonically increasing errors, and the smallest skeletal mechanism that satisfies a user-defined error tolerance over a selected domain of applicability is readily obtained. An additional module based on life-time analysis identifies a set of species that can be modeled accurately by quasi-steady state relations. An application of the reduction procedure is presented for autoignition using a large iso-octane mechanism. The whole process is automatic, is fast, has moderate CPU and memory requirements, and compares favorably to other existing techniques.}
@@ -311,7 +308,6 @@ @article{niemeyer:2011
 year = {2011},
 issn = {0010-2180},
 doi = {10.1016/j.combustflame.2010.12.010},
-url = {https://www.sciencedirect.com/science/article/pii/S0010218010003640},
 author = {Kyle E. Niemeyer and Chih-Jen Sung},
 keywords = {Mechanism reduction, Skeletal mechanism, DRG, DRGEP, Graph search algorithm, Dijkstra’s algorithm},
 abstract = {The importance of graph search algorithm choice to the directed relation graph with error propagation (DRGEP) method is studied by comparing basic and modified depth-first search, basic and R-value-based breadth-first search (RBFS), and Dijkstra’s algorithm. By using each algorithm with DRGEP to produce skeletal mechanisms from a detailed mechanism for n-heptane with randomly-shuffled species order, it is demonstrated that only Dijkstra’s algorithm and RBFS produce results independent of species order. In addition, each algorithm is used with DRGEP to generate skeletal mechanisms for n-heptane covering a comprehensive range of autoignition conditions for pressure, temperature, and equivalence ratio. Dijkstra’s algorithm combined with a coefficient scaling approach is demonstrated to produce the most compact skeletal mechanism with a similar performance compared to larger skeletal mechanisms resulting from the other algorithms. The computational efficiency of each algorithm is also compared by applying the DRGEP method with each search algorithm on the large detailed mechanism for n-alkanes covering n-octane to n-hexadecane with 2115 species and 8157 reactions. Dijkstra’s algorithm implemented with a binary heap priority queue is demonstrated as the most efficient method, with a CPU cost two orders of magnitude less than the other search algorithms.}